Cerebras Systems, a rising star in the tech industry focused on creating what CEO Andrew Feldman calls “the fastest AI hardware for training and inference,” has officially initiated the process to go public.

Originally, the company had intentions of launching an initial public offering in 2024; however, those plans were put on hold due to a federal review regarding an investment from G42, a company based in Abu Dhabi. This led to the withdrawal of their IPO filing. Last year, Cerebras successfully secured a significant $1.1 billion during its Series G funding round, followed by a $1 billion Series H funding round in February, achieving an impressive valuation of $23 billion, as reported by the Wall Street Journal.

In recent developments, Cerebras has forged a partnership with Amazon Web Services, enabling the use of its advanced chips in Amazon’s data centers. Additionally, the company entered into a substantial deal with OpenAI, valued at over $10 billion.

In a conversation with the Wall Street Journal, Feldman expressed confidence in their competitive edge, stating, “Obviously, [Nvidia] didn’t want to lose the fast inference business at OpenAI, and we took that from them.”

According to their filing, Cerebras generated $510 million in revenue for 2025, with a net income of $237.8 million. When adjusting for certain one-time expenses, the adjusted net figure reflected a loss of $75.7 million according to non-GAAP standards.

While the exact amount the company aims to raise through the IPO has yet to be revealed, a spokesperson confirmed that the offering is expected to take place in mid-May.

Cerebras, an AI chip startup, submits IPO application.

A bustling stock exchange floor bathed in warm golden sunlight; at center, a giant AI chip wafer embossed with the Cerebras logo on a sleek pedestal, surrounded by cheering traders in business attire as a digital ticker behind them flashes “CWB ↑” against a bright, optimistic backdrop.

AI chip startup Cerebras files for IPO - TechCrunch

- Cerebras Systems has filed for an IPO, previously delayed due to federal review.  
- The company raised $2.1 billion in funding, achieving a $23 billion valuation.  
- Cerebras partnered with Amazon Web Services and OpenAI for significant commercial deals.  

In Brief

Cerebras Systems, a startup building what CEO Andrew Feldman describes as “the fastest AI hardware for training and inference,” has filed to go public.

The company previously filed for an initial public offering in 2024, but that was delayed due to a federal review of an investment from Abu Dhabi-based G42 and was ultimately withdrawn. Cerebras raised a $1.1 billion Series G last year, followed by a $1 billion Series H in February at a $23 billion valuation, according to the Wall Street Journal.

In recent months, the company announced an agreement with Amazon Web Services to use Cerebras chips in Amazon data centers, as well as a deal with OpenAI reportedly worth more than $10 billion.

In a recent interview with the WSJ, Feldman boasted, “Obviously, [Nvidia] didn’t want to lose the fast inference business at OpenAI, and we took that from them.”

Cerebras brought in $510 million in revenue in 2025, according to the filing, with a net income of $237.8 million (excluding certain one-time items, it was a non-GAAP net loss of $75.7 million).

A company has not disclosed how much it hopes to raise in the IPO. A spokesperson said the offering is planned for mid-May.

AI chip startup Cerebras files for IPO

During a recent address at the Hannover Messe, Chancellor Friedrich Merz of Germany emphasized the necessity for a more flexible regulatory framework for artificial intelligence (AI) in industrial applications within the European Union. He proposed that liberating industrial AI from restrictive regulations is crucial for enhancing productivity compared to consumer AI scenarios.

Merz stated, "I will advocate for a reduction in regulatory constraints associated with AI in the EU, aiming to exempt industrial AI from the current overly stringent regulations that hamper its development." He highlighted AI's potential to drive efficiencies, optimize resource utilization, and significantly lower operational costs.

Germany is keen on positioning itself more competitively in the global AI landscape, challenging the dominance of the United States and China in this transformative technology sector while striving to create high-paying job opportunities. Last month, the German government announced initiatives aimed at quadrupling the nation’s AI data processing capabilities by 2030 to stimulate investments and progress in this field. 

This move aligns with Germany's broader strategy to establish itself as a significant player in the ongoing race for AI expertise and innovation.

Germany's Merz advocates for looser EU regulations on industrial AI.

A modern industrial hall flooded with warm, natural sunlight: sleek robotic arms assembling machinery in the background, in the foreground a confident silhouette of Friedrich Merz at a podium with one German flag behind him. A giant pair of scissors snips through a bright red ribbon labeled “EU Regulations,” while blueprint-style schematics glow softly on nearby screens.

Germany's Merz says industrial AI needs less stringent EU regulation - Reuters

- Chancellor Friedrich Merz advocates for less EU regulation on industrial AI to enhance productivity.  
- He highlights the need for regulatory freedom compared to consumer AI applications for efficiency.  
- Germany aims to compete with the US and China in advancing AI technology and jobs.  

HANOVER, Germany, April 19 (Reuters) - Germany's Chancellor Friedrich Merz said on Sunday that artificial ​intelligence for industrial use will require ‌more regulatory freedom in the European Union than other AI areas such as consumer ​use to boost productivity.

"I will ​push to ease the regulatory burden ⁠in the EU on AI and, ​where possible, to exempt industrial AI ​from the current regulatory straightjacket that is too tight for AI within the European Union," ​Merz said in a speech at ​the annual industrial fair Hannover Messe.

Sign up here.

"AI will contribute ‌to ⁠greater efficiency and productivity, optimised use of resources and, above all, reduced costs," he added.

Germany has been eager ​to catch ​up with ⁠dominant AI players the United States and China in ​a global race to master ​a ⁠transformational technology and attract high-income jobs.

Last month, Berlin to encourage investments to ⁠boost ​AI data processing capacity ​at least fourfold by 2030.

Reporting by Andreas Rinke; Writing ​by Ludwig Burger; Editing by Lisa Shumaker

Our Standards: The Thomson Reuters Trust Principles. , opens new tab

Germany's Merz says industrial AI needs less stringent EU regulation

Donald Trump’s recent self-portrayal as a messianic figure and his contentious remarks about Pope Leo XIV may have significant repercussions for both him and the Republican Party in the upcoming midterm elections. Experts predict that certain Christian groups, which feel alienated by his actions, will play a vital role in pivotal races throughout the United States.

The controversy began when Trump posted an AI-generated image on his Truth Social platform that depicted him bestowing a mystical light upon a distressed man while ethereal figures hovered above. Trump, 79, claimed the image symbolized him in the role of a healer. This interpretation sparked immediate backlash from some Christians, including members of the political right. Douglas Wilson, a prominent Christian nationalist, labeled the image as “blasphemous,” arguing it should be removed without delay. Sean Feucht, a Christian activist collaborating with Trump on a “Worship Tour,” echoed this sentiment, expressing disbelief on social media.

Although Trump's initial post was eventually deleted, skepticism surrounded the sincerity of his apology, particularly after he shared another AI image portraying Jesus embracing him. Kristin Kobes Du Mez, an expert on American evangelicals at Calvin University, noted that, while Trump has a history of behavior that has troubled many within his Christian base, this particular incident felt different. The overt nature of the image seemed to cross a line for some supporters, prompting their reaction.

However, Du Mez cautioned that not all of the backlash may be entirely genuine. She suggested that some Christian leaders felt pressured to publicly distance themselves from Trump without withdrawing their support. Feucht seemed to move past the controversy quickly, as he later reposted Trump’s explanation without critical commentary. Meanwhile, Riley Gaines, a Christian activist, voiced initial disapproval but later praised Trump on social media.

Robert Jones, head of the Public Religion Research Institute, believes that Trump’s backing among white evangelicals and Christian nationalists will likely persist, fueled by their conservative ideologies and a historical context rooted in the South that often aligns with anti-immigrant and anti-Islam sentiments. He indicated that this demographic's existing ties to Trump might overlook his disagreements with religious leaders.

Nonetheless, Jones pointed out that white evangelicals’ influence may not be as critical in the upcoming midterms, particularly in swing states, due to their concentration in solidly red areas. In contrast, white Catholics are expected to have a considerable impact, especially in competitive districts. Data from the PRRI suggests that while a majority of white Catholics supported Trump in the 2020 election, losing their backing could significantly impact election outcomes.

White Catholics represent about 12% of the national population, but their numbers rise in key battleground states, like Pennsylvania at 18%, Wisconsin at 22%, and Michigan, where their presence is notably strong. Jones warned that if Republicans experience a decline in support among white Catholics, it could drastically affect their chances of maintaining power in various midterm elections, emphasizing their potential to sway the results in closely contested races.

Trump gauges support from the religious right during his conflict with the pope and AI Jesus messages

A dimly lit cathedral stage bathed in warm candlelight: Trump stands at a shadowy podium draped with the U.S. flag, his figure backlit by flickering flames. Behind him, a rustic screen displays a glitchy, golden-haloed AI portrait of Jesus composed of drifting binary code. On a nearby stone altar lies a toppled papal miter and a cracked crozier, while silhouetted evangelicals clutch bibles in the darkened pews, the air thick with tension.

Trump tests his luck with the religious right amid feud with pope and AI Jesus posts - The Guardian

- Trump's portrayal of himself as a divine figure has drawn backlash from Christian groups.  
- Support among white evangelicals may remain, but white Catholics hold swing state influence.  
- Loss of support from white Catholics in key states could significantly impact Republican midterms.  

Donald Trump’s depiction of himself as Jesus Christ and recent spat with Pope Leo XIV could come back to bite him and the Republican party in the midterm elections, according to experts, with some newly aggrieved Christian groups set to play an outsized role in key races across the US.

The president’s Trump-as-the-Messiah Truth Social post sparked immediate criticism among some Christians, including some on the right. Trump, 79, said he thought the AI image of him administering an ethereal light to a stricken man’s head as translucent figures descended from the heavens represented him as a doctor.

“Blasphemous,” was the verdict of Douglas Wilson, a self-described Christian nationalist who believes women should not be able to vote, and a confidant of Pete Hegseth, the US defense secretary.

“This should be deleted immediately. There’s no context where this is acceptable,” Sean Feucht, a Christian activist who is partnering with the Trump administration on a “Worship Tour”, posted on social media.

Trump’s post was ultimately taken down, although the level of his apology was brought into question when he then posted an AI image of what appeared to be Jesus Christ cuddling him.

“He did seem to cross a line for some of his Christian supporters,” said Kristin Kobes Du Mez, a professor at Calvin University and an expert on white American evangelicals. She noted that Trump’s supporters have been prepared to put up with plenty of other things.

“Really, from day one, with him bragging on camera that he assaulted women in the Access Hollywood tape to, even just very recently, threatening to annihilate an entire civilization. He is also detaining children, and there are allegations right now related to the Epstein files.

“There is a lot out there that arguably should concern Christian supporters, and the fact that it was this AI-generated image that sparked this outcry is worth pondering. I think it felt like it crossed the line for some because it was just so blatant.”

Yet Du Mez said not all the outrage may have been sincere.

“I sense that there was this kind of performative aspect that enough Christian leaders knew that they needed to be on record saying: ‘We don’t approve of this. But again, that’s something very different from them actually withdrawing their support from him,” she said.

Feucht, the Worship Tour guru, certainly got over it quickly. Within hours he had uncritically reposted Trump’s “doctor” explanation. Riley Gaines, a Christian, anti-trans activist, initially responded to the Trump image: “A little humility would serve him well.” Later that day, Gaines wrote on social media: “I love the President and I’m so grateful he’s in the Oval Office.”

Robert Jones, president and founder of the Public Religion Research Institute (PRRI), said Trump’s support among white evangelicals and Christian nationalists would probably endure.

“They’re just more conservative than Catholics are. They’re heavily concentrated in the south, so they come out of this kind of southern US history, and they’re frankly more motivated by racism. So the anti-immigrant, anti-Islam drumbeat is kind of fuelling the Maga movement. And the flip side of that anti-Islam, anti-immigrant drumbeat is [the right wing’s] positive vision of a white Christian America.”

Trump’s attacks on Pope Leo – the president has called him “wrong on the issues” and, bizarrely, “weak on crime” – also may not be important to that white evangelical base: some have even backed Trump’s attacks on the pontiff. But those Christians are likely to be less important than others in this year’s elections.

“White evangelicals, except for maybe in Georgia and North Carolina, aren’t actually going to help the Republican party in the midterms that much, because most they’re mostly concentrated in fairly safe, red states and red districts. They’re heavily concentrated in the south, where Republicans are going to win anyway,” Jones said.

“If you ask, though, who are the biggest religious groups in the more competitive states and districts, the answer is Catholics.”

White Catholics, in particular, could have a big say in the midterms, PRRI data shows. A majority of them voted for Trump in 2024 – Hispanic Catholics tended to support Harris – meaning a loss in their support could influence competitive races.

And unfortunately for the Republican party, white Catholics are overrepresented in several swing states, where there are seats which fluctuate between Republican and Democrat. White Catholics make up 12% of the nation as a whole, according to PRRI, but in Pennsylvania they make up 18% of the population; in Wisconsin 22%, and they overindex in Michigan, too.

“In these more competitive districts and swing states, if he loses 10 points among white evangelicals, he and Republicans might be able to weather that. If they lose 10 points among white Catholics, that’s going to be game over in many elections in the midterms,” Jones said.

Trump tests his luck with the religious right amid feud with pope and AI Jesus posts

Singaporean artist Wenhui Lim, known as Niceaunties, stands at the captivating crossroads of digital artistry, speculative storytelling, and women’s perspectives. Instead of merely using artificial intelligence as a shortcut, she embraces it as a generative partner to craft immersive and surreal realms that challenge societal expectations of women, particularly the often-overlooked archetype of the "auntie." Her artistic journey is deeply influenced by mundane life, travel experiences, and the whimsical oddities encountered in daily routines. The creative process unfolds slowly, beginning with fragments of dreams, sketches, and narrative concepts that take shape through a collaborative dialogue with AI tools like Midjourney and Dall-E 3. The resulting unexpected visuals help steer her storytelling in new and innovative directions.

The images crafted during this process undergo meticulous refinement, animation, and assembly into dynamic cinematic sequences using a range of advanced platforms. The outcome navigates a space between fantasy and familiarity, featuring bizarre combinations of food, animals, and human figures that blend seamlessly within a dreamlike, kawaii-inspired aesthetic. Her expanding repertoire of AI-generated films showcases not only technical skill but a robust conceptual framework, highlighting the exciting potential inherent in the collaboration between humans and AI. This co-authorship allows for a flexible approach to control, where meaning is formed through experimentation, spontaneity, and playful interaction. 

In terms of her creative workflow, Niceaunties elaborates on her extensive process, which ranges from early ideas to the completion of the final piece. She explains that her initial steps involve simmering ideas in her mind before translating them into visual forms with AI tools. Engaging in initial brainstorming through written notes or sketches and establishing a broad narrative framework sets the stage for her work. While utilizing AI programs, the organic and unanticipated results frequently lead to fresh ideas, guiding the evolution of the narrative. After generating visuals, she enhances them with Magnific.ai and animates sequences using AI image-to-video software, such as Runwayml and PIKAlabs. Editing culminates in her final video, often requiring her to revisit earlier steps to supplement the storyline and integrate AI-generated music and sound.

The duration of creating a single video can vary significantly, spanning from a single day to several weeks, heavily influenced by the flow of inspiration drawn from daily life, travel, and experiences. Her artistic approach incorporates inspirations from surrealism, fantasy, and kawaii culture, resulting in visually striking pieces that showcase unexpected collages of seemingly unrelated objects—like sushi and legs—juxtaposed in dreamlike normality. The surreal combinations and cute elements infuse joy into her work, creating an engaging atmosphere.

Regarding the emerging AI art community, Niceaunties reflects on the supportive nature of this new environment, which is comprised of diverse individuals, from traditional artists to non-artists who are nevertheless creatively inclined. 

Her recent works include a featured film on a prominent billboard at 9157 Sunset Boulevard in Los Angeles and an exhibition titled “Collaborations with the Artificial Self” at Art Basel. The film “Aunties on Sunset” runs for three and a half minutes, depicting a surreal outing of aunties in Los Angeles and West Hollywood. In her group show, she presented three AI-generated videos created, animated, and edited using Dall-E 3, Magnific.ai, Runwayml, and Suno.ai.

Among her creations, “Dessert Desert” narrates the story of an auntie who wakes up confused and hungry in a vast sandy expanse. Her quest for food leads her to unearth various creatures and delightful foods from the sand, culminating in a scene where she enjoys dessert in a kitchen setting. This dream transforms to reveal a commentary on the inner landscapes of our minds, highlighting how a nuanced shift in perspective can enhance our enjoyment of life.

In “Auntie’s Time Machine (ATM),” the protagonist yearns for a beloved cake from the past. To relive that experience, she uses her robotic multipurpose camel, ATM, to embark on a nostalgic adventure through dessert-laden landscapes. Encountering humorous obstacles along the way, she ultimately discovers that her journey was more meaningful than the cake itself, exploring the notion that the journey can sometimes overshadow the destination.

“Camelot” tells the tale of Auntie and her robot camel running out of fuel in a whimsical desert setting. Their encounter with mobile shop camels leads them to the magical oasis called Camelot, which transcends the simple notion of a resting place for camels, evolving into a sanctuary for various burden-bearing creatures. This segment creatively weaves pop culture references and wordplay into the narrative, enhancing the playful essence of her art.

Niceaunties Redefines Womanhood with AI Dreamscapes

A sunlit art studio loft with tall windows flooding warm natural light onto wooden floors. Centered is a life-sized digital canvas alive with swirling pastel petals, floating cityscapes and lunar rings—an ethereal AI dreamscape reimagining womanhood. A confident adult woman in profile reaches toward the scene. The Niceaunties logo sits prominently on the canvas frame, tying the visionary artwork to the brand.

Niceaunties Reimagines Womanhood Through AI Dreamscapes - Forbes

- Wenhui Lim, known as Niceaunties, uses AI to create surreal feminist narratives in her art.  
- Her creative process combines dreams, sketches, and AI tools like Midjourney and Dall-E 3.  
- Recent works include films showcasing aunties exploring fantastical themes and surreal landscapes.  

Niceaunties, Chapter No. 3, Factory #0201, 2024 Photo courtesy of Niceaunties

Singaporean artist Wenhui Lim, who works under the moniker Niceaunties, occupies a distinctive position at the intersection of digital art, speculative storytelling and female commentary, using artificial intelligence not as a shortcut but as a generative collaborator in constructing richly layered, surreal worlds that subvert expectations placed on women—particularly the often-overlooked figure of the “auntie”. Rooted in observations of everyday life, travel and the quiet absurdities of daily routines, her practice begins with a slow incubation of ideas—dream fragments, sketches and narrative outlines—that evolve through an iterative dialog with AI tools such as Midjourney and Dall-E 3, where unexpected visual outputs actively shape the direction of her storytelling.

These images are subsequently refined, animated and assembled into cinematic sequences using a suite of emerging platforms, resulting in works that oscillate between fantasy and familiarity, often populated by uncanny juxtapositions—food, animals and human forms coexisting in dreamlike logic informed by kawaii esthetics. Her expanding body of AI-driven films demonstrates not only a technical fluency but also a conceptual rigor, foregrounding the creative potential of human-AI partnerships as spaces of co-authorship, where control is deliberately porous and meaning emerges through negotiation, chance and play.

Take me through the different steps of your creative and production process, from conceptualization to the final piece. Tell me about the techniques, equipment, software, AI programs, materials and formats you use.

I spend a lot of time dreaming and marinating ideas before I prompt text to images in collaboration with AI in programs like Midjourney and Dall-E 3. Sometimes I write down or sketch ideas I dream about, making a list of things I have yet to create. I always start with a broad story outline and a central idea. While creating images in AI programs, new ideas emerge from the organic prompting process or unexpected details generated by the machine, inspiring new directions. The narrative develops through this process. I upscale the images in Magnific.ai before animating them using image-to-video AI programs like Runwayml and PIKAlabs, and edit the sequences into a final video. Often, I return to the first step of creating images to fill in the story before completing the work, incorporating AI music creation and sound editing.

How long does it take you to create a single work, and how do you come up with the different storylines and scenarios for your videos?

Creating a video takes anything from a day to a few weeks, depending on the flow of inspiration. I am inspired by everyday life, travel and events.

How do your sources of inspiration—surrealism, fantasy and kawaii culture—influence your artworks stylistically?

My art tends to be a collage of things and objects that do not typically exist together, such as sushi and legs. I also enjoy the feeling of strange things seeming normal in dreams, only to appear strange upon waking. I express this surreal juxtaposition in my work. Kawaii elements like toys and animals make me happy and frequently appear in my artworks.

Niceaunties, Camelot, More than a Parking Lot for Camels Photo courtesy of NIceaunties

What are your thoughts on the AI art community and how do you collaborate with other AI artists?

The AI art community is very new and supportive, comprising individuals from all walks of life, including some traditional artists and many creatives who do not work as artists.

Describe your film on the billboard at 9157 Sunset Boulevard in Los Angeles and your exhibition “Collaborations with the Artificial Self” at Art Basel.

“Aunties on Sunset” on the billboard at 9157 Sunset Boulevard featured a 3.5-minute film of aunties’ surreal visit to LA and West Hollywood. Collaborations with the Artificial Self” was a group show with many prominent AI artists through the timeline of AI art. I exhibited three AI videos, which I created, animated and edited in the programs Dall-E 3, Magnific.ai, Runwayml and Suno.ai.

“Dessert Desert” begins with an auntie waking up confused and hungry in a vast desert. Driven by hunger, she starts digging into the sand, and as she digs, various creatures and bits of food begin to appear, eventually revealing cream and giant strawberries. The scene then transitions to a kitchen, where the auntie is enjoying her dessert of crumble, strawberries and cream. The earlier scene is revealed to be a dream induced by food fatigue. The film reflects on our inner worlds and imagined bleakness, suggesting that a small shift in perspective—much like the wordplay between “dessert” and “desert”—can lead to a much sweeter enjoyment of life.

“Auntie’s Time Machine (ATM)” tells how a long time ago, auntie enjoyed a lovely cake. Missing the taste, she decided to travel back in time using her robotic multipurpose camel, ATM, to search for that flavor. Auntie and her machine journeyed far and wide across landscapes of dessert-strewn deserts. They occasionally stopped for a cold beer provided by her robot camel and even faced terror from some cat robbers. Eventually, she found the cake at an oasis, but by then, she was no longer hungry, content with the wonderful memories and adventures she had experienced. Loosely inspired by The Alchemist, this tale reflects on personal journeys and the notion that end goals are not always the most important aspect. This story is part of the “Dessert Desert” series, where Niceaunties explores surprises in barren, desolate landscapes.

“Camelot” relates how while traveling across the dessert desert, Auntie’s robot camel, Auntie’s Time Machine (ATM), ran out of gas. They encountered some mobile shop camels who advised them to refill at the legendary oasis, the Camelot. Camelot is revealed to be more than just a parking lot for camels and lost travelers; it’s a haven for all creatures of burden, known only to aunties and various camels in the multiverse. The shop catalog showcases camels of all kinds, including the cigarettes camel, camels of the round table, a “camelflage” and a very buff camel bodyguard. In this episode, references to pop culture and wordplay are used extensively.

Niceaunties Reimagines Womanhood Through AI Dreamscapes

In the late 1980s, economist Robert Solow, a Nobel laureate, remarked on the stagnation of productivity growth despite the advancements in technology during the Information Age, such as transistors and microprocessors. He noted that, contrary to expectations that these innovations would enhance workplace efficiency, productivity actually declined—from 2.9% between 1948 and 1973 to just 1.1% after that. Instead of driving a productivity boom, the new computing technologies often overwhelmed users with excessive information, producing detailed reports and volumes of paper rather than the efficiency gains anticipated. This unexpected phenomenon became known as Solow’s productivity paradox. As he famously stated, "You can see the computer age everywhere but in the productivity statistics," in a New York Times Book Review article in 1987.

Fast forward to today, and there seems to be a resurgence of similar concerns regarding artificial intelligence (AI) within corporate America. Recent research, including an analysis by the Financial Times, reveals that while many C-suite executives are optimistic about AI's potential, that optimism is not translating into significant productivity improvements. Although 374 firms in the S&P 500 mentioned AI positively in their earnings calls, an extensive study of 6,000 executives found that a large percentage see minimal impact from AI on their workplaces. Roughly two-thirds utilize AI, but their average use amounts to just 1.5 hours per week, with one-quarter not employing AI at all in their operations. The majority of companies reported no changes in productivity or employment linked to AI over the past three years.

Yet, executives are hopeful for future advancements, projecting an anticipated productivity rise of 1.4% and an output increase of 0.8% within the next three years. Interestingly, while businesses expect a slight reduction in jobs, employees surveyed indicated an increase in employment prospects.

Research from MIT in 2023 suggested that AI could boost individual worker performance by nearly 40%, but mixed results from the data have left economists questioning when or if AI will deliver a return on investment, especially as corporate expenditures on AI surpass $250 billion in 2024. Apollo's chief economist, Torsten Slok, echoed Solow's sentiments when he stated that today, like decades ago, AI is not evident in important economic metrics like employment or productivity data. He noted that aside from a few high-profile tech companies, there are no discernible improvements in profit margins or earnings forecasts linked to AI deployment.

Diverse studies have painted conflicting pictures regarding AI’s effectiveness. Reports indicate some slight productivity increases, such as a 1.9% rise noted by the Federal Reserve Bank of St. Louis, but a more recent MIT study reported only a modest 0.5% gain over ten years. Nobel laureate Daron Acemoglu acknowledged that while an increase is positive, it pales in comparison to industry expectations.

Further complicating the discourse is a study by ManpowerGroup, which revealed that while regular AI usage among workers surged by 13% in 2025, confidence in the technology dropped by 18%, highlighting ongoing skepticism about its real benefits. Moreover, a Boston Consulting Group study suggests that using too many AI tools could lead to decreased productivity, as employees reported feeling overwhelmed and making more mistakes when juggling numerous applications.

IBM’s chief human resources officer remarked on the company's plan to increase its workforce, pointing out that while AI can automate tasks, it could also hinder the progression of future leaders by displacing entry-level positions.

Could this trend change in the future? Historical trends show that the IT advancements of the 1970s and 1980s eventually led to significant productivity booms in the 1990s and early 2000s. Erik Brynjolfsson, director of Stanford University's Digital Economy Lab, suggested that we may witness such a turnaround with current AI implementations, observing a notable GDP increase and a recent productivity uptick attributed to firms beginning to effectively leverage AI technology. This is reminiscent of past trends where job growth and GDP growth became less correlated due to automation.

Interestingly, a Stanford study noted hidden productivity gains, revealing that AI has significantly improved efficiency in everyday online tasks, although the time saved often shifts towards leisure rather than skill advancement.

Slok likened the future trajectory of AI productivity to a “J-curve,” where initial underperformance could be followed by significant improvements, depending on how effectively organizations integrate AI into their operations. Unlike previous IT innovations that had lengthy development cycles, AI tools are quickly accessible due to competitive forces, making its adoption and execution in diverse industries crucial for maximizing its economic potential. Ultimately, the value will not stem solely from the technology itself but from innovative applications across sectors.

Many executives are missing out on the AI productivity surge, echoing the paradox of the IT era for economists.

A shadow-filled corporate boardroom at night, executives (adults only) slumped around a long glossy table, a large digital screen at the head displaying a flatline AI productivity graph in muted red, the wall behind them ghosting the faint outline of a 1980s mainframe, low warm tungsten light casting dramatic shadows.

Thousands of CEOs admit AI had no impact on employment or productivity—and it has economists resurrecting a paradox from 40 years ago - Fortune

- Robert Solow’s productivity paradox observed that tech advancements didn't lead to productivity gains.  
- Most executives report minimal impact of AI on productivity and employment in their firms.  
- Future productivity gains from AI may depend on effective implementation across different sectors.  

In 1987, economist and Nobel laureate Robert Solow made a stark observation about the stalling evolution of the Information Age: Following the advent of transistors, microprocessors, integrated circuits, and memory chips of the 1960s, economists and companies expected these new technologies to disrupt workplaces and result in a surge of productivity. Instead, productivity growth slowed, dropping from 2.9% from 1948 to 1973, to 1.1% after 1973.

Newfangled computers were actually at times producing too much information, generating agonizingly detailed reports and printing them on reams of paper. What had promised to be a boom to workplace productivity was for several years a bust. This unexpected outcome became known as Solow’s productivity paradox, thanks to the economist’s observation of the phenomenon.

“You can see the computer age everywhere but in the productivity statistics,” Solow wrote in a New York Times Book Review article in 1987.

Data on how C-suite executives are—or aren’t—using AI shows history is repeating itself, complicating the similar promises economists and Big Tech founders made about the technology’s impact on the workplace and economy. Despite 374 companies in the S&P 500 mentioning AI in earnings calls—most of which said the technology’s implementation in the firm was entirely positive—according to a Financial Times analysis from September 2024 to 2025, those positive adoptions aren’t being reflected in broader productivity gains.

A study published in February by the National Bureau of Economic Research found that among 6,000 CEOs, chief financial officers, and other executives from firms who responded to various business outlook surveys in the U.S., U.K., Germany, and Australia, the vast majority see little impact from AI on their operations. While about two-thirds of executives reported using AI, that usage amounted to only about 1.5 hours per week, and 25% of respondents reported not using AI in the workplace at all. Nearly 90% of firms said AI has had no impact on employment or productivity over the last three years, the research noted.

However, firms’ expectations of AI’s workplace and economic impact remained substantial: Executives also forecast AI will increase productivity by 1.4% and increase output by 0.8% over the next three years. While firms expected a 0.7% cut to employment over this time period, individual employees surveyed saw a 0.5% increase in employment.

Is AI actually making people more productive?

In 2023, MIT researchers claimed AI implementation could increase a worker’s performance by nearly 40% compared to workers who didn’t use the technology. But emerging data failing to show these promised productivity gains has led economists to wonder when—or if—AI will offer a return on corporate investments, which swelled to more than $250 billion in 2024.

“AI is everywhere except in the incoming macroeconomic data,” Apollo chief economist Torsten Slok wrote in a blog post, invoking Solow’s observation from nearly 40 years ago. “Today, you don’t see AI in the employment data, productivity data, or inflation data.”

Slok added that outside of the Magnificent Seven, there are “no signs of AI in profit margins or earnings expectations.”

Slok cited a slew of academic studies on AI and productivity, painting a contradictory picture about the utility of the technology. Last November, the Federal Reserve Bank of St. Louis published in its State of Generative AI Adoption report that it observed a 1.9% increase in excess cumulative productivity growth since the late-2022 introduction of ChatGPT. A 2024 MIT study, however, found a more modest 0.5% increase in productivity over the next decade.

“I don’t think we should belittle 0.5% in 10 years. That’s better than zero,” study author and Nobel laureate Daron Acemoglu said at the time. “But it’s just disappointing relative to the promises that people in the industry and in tech journalism are making.”

Other emerging research can offer reasons why: Workforce solutions firm ManpowerGroup’s 2026 Global Talent Barometer found that across nearly 14,000 workers in 19 countries, workers’ regular AI use increased 13% in 2025, but confidence in the technology’s utility plummeted 18%, indicating persistent distrust.

AI adoption can even be counterproductive at a certain point, according to a study conducted by Boston Consulting Group, leading to “AI brain fry.” In a survey of 1,488 full-time U.S.-based workers, respondents reported increased productivity when using three or fewer AI tools, but self-reported productivity plummeted when respondents used four or more tools, with workers saying they felt brain fog or made more small mistakes as a result of technology overuse.

Nickle LaMoreaux, IBM’s chief human resources officer, said this year the tech giant would triple its number of young hires, suggesting that despite AI’s ability to automate some of the required tasks, displacing entry-level workers would create a dearth of middle managers down the line, endangering the company’s leadership pipeline.

What could reverse AI’s productivity pattern?

To be sure, this productivity pattern could reverse. The IT boom of the 1970s and ’80s eventually gave way to a surge of productivity in the 1990s and early 2000s, including a 1.5% increase in productivity growth from 1995 to 2005 following decades of slump.

Economist and Stanford University’s Digital Economy Lab director Erik Brynjolfsson noted in a Financial Times op-ed the trend may already be reversing. He observed that fourth-quarter GDP was tracking up 3.7%, despite last week’s jobs report revising down job gains to just 181,000, suggesting a productivity surge. His own analysis indicated a U.S. productivity jump of 2.7% last year, which he attributed to a transition from AI investment to reaping the benefits of the technology. Former Pimco CEO and economist Mohamed El-Erian also noted job growth and GDP growth continuing to decouple as a result in part of continued AI adoption, a similar phenomenon that occurred in the 1990s with office automation.

Some productivity increases may be hiding in plain sight. A study led by the Stanford Institute for Economic Policy Research found using internet browsing data from 200,000 U.S. households that generative AI increased the efficiency of online tasks like job hunting, travel planning, or shopping from between 76% and 176%. However, researchers found the time AI users saved on chores was spent hanging out with friends or watching television, as opposed to spent on work on new skills development.

Slok saw the future impact of AI as potentially resembling a “J-curve” of an initial slowdown in performance and results, followed by an exponential surge. He said whether AI’s productivity gains would follow this pattern would depend on the value created by AI.

So far, AI’s path has already diverged from its IT predecessor. Slok noted in the 1980s, an innovator in the IT space had monopoly pricing power until competitors could create similar products. Today, however, AI tools are readily accessible as a result of “fierce competition” between large language model-buildings driving down prices.

Therefore, Slok posited, the future of AI productivity would depend on companies’ interest in taking advantage of the technology and continuing to incorporate it into their workplaces. “In other words, from a macro perspective, the value creation is not the product,” Slok said, “but how generative AI is used and implemented in different sectors in the economy.”

A version of this story was published on Fortune.com on Feb. 17, 2026.

More on AI and productivity:

Thousands of executives aren't seeing AI productivity boom, reminding economists of IT-era paradox

Despite the excitement from executives about the potential of artificial intelligence agents to handle an array of office tasks with relentless efficiency, the technology itself remains fragile and could become a financial liability.

This week, at two distinct gatherings in Silicon Valley, leaders and engineers explored both the thrill and the hurdles tied to AI agents. Kevin McGrath, CEO of the AI startup Meibel, highlighted a significant issue in the field: the common misconception that every task must go through a large language model (LLM). 

He cautioned against mindlessly allocating resources to AI systems, saying, “Just give all of your tokens and all of your money to an AI Claw bot that will just waste millions and millions of tokens,” urging companies to be cautious and strategic about which tasks are appropriate for AI utilization.

The surge of interest in AI agents has been fueled by innovations like OpenClaw, a platform enabling developers to utilize different AI models for building and overseeing teams of digital assistants, which have been touted as the next big breakthrough in the tech world. Nvidia's CEO Jensen Huang even characterized this development as the evolution beyond ChatGPT during a conversation with CNBC's Jim Cramer in March.

However, discussions at the Generative AI and Agentic AI Summit in San Jose made it clear that deploying and managing AI agents is fraught with complexities. A session featuring Google software engineer Deep Shah emphasized the need for new methodologies aimed at controlling the operational costs associated with numerous AI agents.

Running these digital assistants involves expenses, and a poorly structured system to oversee their activities can result in unexpected financial drains rather than savings. Shah elaborated, “If you think of a machine learning system or any multi-agent system, there are multiple challenges you will find when you try to deploy that system at scale. The first one is the inference cost.”

Silicon Valley's AI agents face issues: Inefficient token usage and disorderly systems

Dark-themed, moody server room carved into Silicon Valley hills, the neon “Silicon Valley” sign flickering red through a cracked window. In the foreground, a glitching AI terminal with a shattered screen spews golden tokens that disintegrate mid-air. Tangled cables spark around chaotic server racks under red emergency lights, cinematic atmosphere, natural lighting from emergency lamps.

SiIicon Valley's AI agent hiccups: Wasted tokens and 'chaotic' systems - CNBC

- AI agents are seen as innovative but can lead to significant operational costs.  
- Experts highlight the need for careful task selection for AI implementation success.  
- Managing AI agents at scale presents multiple challenges, particularly with inference costs.  

Despite the C-suite's enthusiasm over artificial intelligence agents that can plow through office tasks like never-sleeping interns, the underlying technology is still rickety and a potential cost-sucker.

That much was clear this week during two separate events held in Silicon Valley, during which executives and engineers discussed the current excitement and challenges involving AI agents.

Kevin McGrath, the CEO of the AI startup Meibel said during a session that "the biggest problem that we're working with in AI right now," involves the misguided idea that everything needs to be processed by a large language model, or LLM.

"Just give all of your tokens and all of your money to an AI Claw bot that will just waste millions and millions of tokens," McGrath said, before explaining how companies need to be more deliberate when deciding which tasks are best suited for AI agents.

Since the recent rise of OpenClaw, a so-called "harness" that lets developers use various AI models to create and manage fleets of digital assistants, the tech industry has been pushing AI agents as the next big thing.

Nvidia CEO Jensen Huang told CNBC's Jim Cramer in March that it "is definitely the next ChatGPT."

But on Wednesday at the Generative AI and Agentic AI Summit in San Jose, technical staff from companies like Google and its DeepMind AI unit, Amazon , Microsoft and Meta revealed that creating and operating AI agents is not an easy task.

One session led by Google software engineer Deep Shah focused on new techniques intended to help manage the operational costs of running tons of AI agents.

It costs money to run AI agents, and a poorly designed and maintained system to monitor those digital assistants and their actions could potentially end up burning cash instead of saving it.

"If you think of a machine learning system or any multi-agent system, there are multiple challenges you will find when you try to deploy that system at scale," Shah said. "The first one is the inference cost."

SiIicon Valley's AI agent hiccups: Wasted tokens and 'chaotic' systems

In the suburbs of Salt Lake City, a group of Republican activists convened to discuss this year's elections, delving into familiar conservative hot topics such as the state's dwindling water resources, concerns over immigration fraud, and various conspiracy theories. However, Doug Fiefia, a state representative aiming to become a state senator, chose to introduce a different discussion point: artificial intelligence. Fiefia, a former Google employee, is leveraging his tech background to center his campaign around the regulation of this rapidly evolving industry. 

"I know it might sound like I'm fixated on this," Fiefia remarked. "But this technology is arriving, and it will be our greatest challenge."

His focus on AI has put him at odds with the Trump administration, which intervened this year to block his proposal requiring companies to establish child safety measures in AI technologies. The White House is advocating for a uniform national standard on AI, citing concerns that a fragmented regulatory approach could stifle American innovation in the competitive landscape against China.

With little advancement at the federal level, state lawmakers have been left to tackle the implications of AI technology on their own. For instance, Florida's Governor Ron DeSantis has added AI regulation to the agenda of an upcoming special legislative session, while New York has mandated that major AI developers report incidents that pose a danger to the public.

Currently, there are over 1,000 proposals across various state legislatures that address AI, showcasing the widespread anxiety surrounding this technology.

"There's a lot of uncertainty," noted Brett Young, a structural engineer who participated in the meeting with Fiefia. "Are we justified in our fears, or is it overblown and potentially beneficial?"

At the state level, the Trump administration has sought to curb local AI regulations through executive orders, which include legal threats and funding penalties aimed at deterring new laws. Recently, the White House unveiled a framework for possible congressional legislation that would preempt state laws deemed overly restrictive, while still permitting some regulations that protect children and copyright.

Despite these efforts, the number of proposals being considered at the state level continues to grow. Popular initiatives include requiring chatbots to identify as non-human and prohibiting the use of AI to create non-consensual pornography.

Craig Albright, senior vice president for government relations at the Business Software Alliance, noted that state lawmakers are closely monitoring federal actions and feel a pressing need to act due to dissatisfaction with the federal response.

A recent Quinnipiac poll revealed that about 80% of Americans express concern regarding AI, with many believing the government is falling short in regulating the technology, including a significant number of both Democrats and Republicans who desire more intervention.

California and New York, both led by Democratic majorities, have implemented the most significant regulatory initiatives, focusing on transparency about catastrophic risks posed by AI, such as malfunctioning systems in critical sectors like nuclear power.

Interestingly, there is also momentum for regulation in Republican-led states. DeSantis advocated for a bill that would require parental controls for minors using AI and prevent the unauthorized use of individuals' likenesses, although this measure stumbled in the House after passing the Senate with strong support.

Fiefia is part of a growing network of former tech professionals who have transitioned into public office, aiming to meet the demand for tighter oversight on AI technologies. He co-leads an AI task force under the Future Caucus with Monique Priestley from Vermont, who shares similar tech-driven legislative goals.

"We collaborate through video calls and group chats to brainstorm proposals and navigate lobbying challenges against our initiatives," Priestley explained, highlighting the significant influence lobbyists have in legislative debates.

Currently, Alex Bores, a former data scientist turned political candidate, faces industry pushback as he competes in the Democratic primary for a congressional seat. A pro-AI committee has heavily criticized him, suggesting that tech companies are using him as an example to dissuade regulatory efforts.

Fiefia has not yet garnered the same level of attention as Bores in his push for a Senate seat. His district is suburban in nature, characterized by its picturesque views and family-friendly atmosphere.

Having grown up in Utah as the son of Tongan immigrants, Fiefia worked at Google where he helped implement early AI models. His experiences there left him skeptical of the tech industry's priorities. 

"Big Tech is primarily focused on profits rather than the welfare of humanity," Fiefia stated, contrasting his current role at a Utah cloud computing company with his previous tech endeavors.

Fiefia's legislative efforts gained traction when his proposal received unanimous approval from a House committee, but it ultimately faced an uphill battle due to opposition from the Trump administration, which deemed it "unfixable."

Fiefia's opponent in the primary, Senator Daniel McCay, expressed relief that the proposal did not advance, arguing against the notion that new technologies are inherently detrimental to society. 

During the discussion among activists, Fiefia was confronted with questions about defying federal authority, asserting that it is vital to prioritize states' rights, particularly when it involves crucial issues like AI.

"The Trump administration promotes zero regulations on AI," he elaborated. "While I align with many of Trump's views, I fundamentally disagree with him on this topic."

Trump aims to prevent states from controlling AI, but this Utah Republican isn't paying attention.

A dramatic, dimly lit corridor inside the U.S. Capitol, warm spotlight cutting through the shadows. On the left, a menacing silhouette of Trump—his unmistakable hairlit by the light—points angrily at a giant parchment stamped “AI Regulation.” On the right, a defiant Utah congressman stands with his back turned, arms crossed, a small Utah flag pin on his lapel. Behind them, a large U.S. flag drapes the marble walls, and faint circuit-board patterns glow etched into the polished floor.

Trump wants to stop states from regulating AI. This Utah Republican isn’t listening - AP News

- Doug Fiefia, a GOP candidate, emphasizes AI regulation amid concerns over its impact.  
- Trump administration opposes state-level AI regulations, advocating for a unified national standard.  
- Growing public anxiety over AI leads to numerous legislative proposals across various states.

Add AP News as your preferred source to see more of our stories on Google.

Add AP News on Google Add AP News as your preferred source to see more of our stories on Google. Share

RIVERTON, Utah (AP) — When a dozen Republican activists gathered on a back deck in the Salt Lake City suburbs to talk about this year’s elections, the conversation cycled through all the staples of conservative chatter in Utah such as dwindling water supplies, illegal immigrant fraud and chemtrail conspiracy theories.

But Doug Fiefia, a state representative running to be a state senator, wanted to start with something else — artificial intelligence. Fiefia used to work at Google and, like several other tech employees who have gone into politics, he has made regulating the industry a centerpiece of his campaign.

“I know it sounds like ‘Doug, this is all you talk about,”’ Fiefia said. “That’s because it’s coming, it’s here and it’s going to be our biggest fight.”

Fiefia’s focus has put him on a collision course with President Donald Trump’s administration, which this year helped block his state proposal requiring companies to include child safety protocols. The White House wants a single national standard for artificial intelligence, arguing that a patchwork of excessive regulation could handicap American innovation in a global competition with China.

But with no progress in Congress, it has been state lawmakers struggling to address concerns about a technology that is poised to reshape the economy. In Florida, Republican Gov. Ron DeSantis added the issue to a special legislative session that he is convening later this month. Democratic-controlled New York last year required major AI developers to report dangerous incidents to the state.

All told, there are more than 1,000 state legislative proposals addressing AI, a reflection of the uneasiness that has seeped through the country.

“None of us are really sure,” said Brett Young, a structural engineer who attended the backyard event with Fiefia. “Is this something we should be scared about, or is it no so big a deal and it’ll enhance our lives?”

Read More

Pressure in the states

Trump has routinely tried to stamp out state-level AI policies, and he issued an executive order that included legal threats and funding penalties to deter new regulations.

The White House recently released a framework for potential congressional legislation that calls for preempting state laws considered “too burdensome” but would allow some rules to protect children and copyright material.

None of these steps has eased the number of proposals in state capitals. Popular ideas include forcing chatbots to remind users they are not human and barring the use of AI to make nonconsensual pornography, which includes replacing or removing clothing from photos that are posted online.

“There’s a lot of state lawmakers looking at what the federal government is doing and saying, ‘We want to take action because we’re not satisfied,’” said Craig Albright, senior vice president for government relations for the Business Software Alliance, which represents software companies.

Sign up for Morning Wire: Our flagship newsletter breaks down the biggest headlines of the day. Email address Sign up By checking this box, you agree to AP's Terms of Use and acknowledge that AP may collect and use your data pursuant to our Privacy Policy

About 8 in 10 people in the United States said they were “concerned” or “very concerned” about AI in a Quinnipiac poll last month, with about three-quarters saying government is not doing enough to regulate the technology. Roughly 9 in 10 Democrats and 6 in 10 Republicans wanted more government involvement.

The most significant regulations have passed in California and New York, solidly Democratic states. The provisions focus on disclosure of catastrophic risk, such as the AI-controlled meltdown of nuclear plants or AI models refusing to heed human direction.

But there is pressure in Republican-led states, too.

DeSantis pushed a bill to implement parental controls for minors using AI and to prohibit systems from using anyone’s likeness without permission. It fell short in the state House after overwhelmingly passing the state Senate. AI bills in Republican-controlled Louisiana and Missouri have stalled out because of Trump administration resistance.

‘An army of full-time lobbyists’

Fiefia is part of a loose network of former tech employees turned state lawmakers trying to meet the demand for stronger regulations. He co-chairs the AI task force of the Future Caucus, a network of younger state lawmakers, with Monique Priestley, a Vermont Democrat who also has worked in tech.

Priestley said the group uses video conferences and group chats to share ideas for new proposals and deal with lobbyists who oppose their bills. She said that 166 of her state’s 482 registered lobbyists weighed in on her data privacy bill last year, which was ultimately vetoed by the governor.

“It’s like you’re running around against an army of full-time lobbyists,” said Priestley. Like many state lawmakers, she works a separate, full-time job.

Alex Bores, a former data scientist at the tech firm Palantir who quit after it signed a deal to help the first Trump administration with immigration enforcement, is also a member of the AI task force. A Democrat, Bores wrote the New York bill that was signed into law last year.

Now Bores is competing in the crowded Democratic primary to replace retiring U.S. Rep. Jerrold Nadler representing the east side of Manhattan and parts of Queen and Brooklyn in Congress, and he is facing payback from the industry. A pro-AI campaign committee has spent $2.3 million against his candidacy.

Bores said tech companies are trying to make an example of him to scare off more regulation at the state and federal level.

“It’s one reasons it’s so important for me to win this race is because, if I don’t, that intimidation they’re trying on Congress will be successful,” he said. Bores’ competitors in the June 23 primary include Jack Schlossberg, the grandson of former President John F. Kennedy, and George Conway, a former Republican who has become one of Trump’s chief antagonists on social media.

From Google to politics

Fiefia has not attracted the sort of attention as Bores as he tries to move to the state Senate after a single session in the House. The subdivisions and shopping centers of his district are sandwiched between Utah’s jagged mountain ranges and the cul de sacs are crammed with children on bikes and scooters.

The son of Tongan immigrants, Fiefia grew up in Utah but moved to Silicon Valley, where he worked as a salesperson for Google.

Fiefia rose to manage a team working with companies on the implementation of Google’s early AI model and was disturbed by what he saw.

“What I realized is Big Tech cares about their bottom line, and they were worried about making money, not doing right for the human race,” said Fiefia, who now works at a Utah-based cloud computing and AI company.

Fiefia’s legislation was unanimously passed by a House committee this year, but the Trump administration sent a letter to the Senate saying that the measure was “unfixable.” The measure quickly died.

Daniel McCay, the state senator who Fiefia is challenging in the primary, said he thinks that was a good thing.

“I’ve been around long enough to recognize the invention of fire, the wheel, cars and the internet did not ruin society and I’m very skeptical of anyone trying to scare society into regulations,” McCay said in an interview.

He noted that the bill went beyond child safety, including whistleblower protection for AI workers and public disclosure of risks.

“It would have driven Utah out of the AI innovation business,” McCay said.

At the cottage meeting — the Utah term for a small gathering at someone’s home to discuss important issues — Fiefia faced several tech-related questions from the crowd.

Asked about defying the Trump administration, Fiefia said it was especially important to stand up for states’ rights when a fellow Republican was in power to demonstrate the principles involved.

“The Trump administration is, ‘We want zero regulations on AI,’” Fiefia said. “I think that’s wrong. I agree with a lot of what Trump says on taxes. I disagree with him on this.”

Trump wants to stop states from regulating AI. This Utah Republican isn’t listening

Environmentalist Paul Ehrlich's predictions from the 1960s about overpopulation leading to resource depletion and widespread starvation have long influenced discussions about the planet’s future. His concerns about humanity’s excessive consumption even deterred some young people from starting families. 

However, it seems Ehrlich was mistaken. As we shift away from fears of overpopulation, we now face a serious demographic crisis: the issue isn't too many births, but rather too few. This decline in birth rates is significant and pressing.

Recent government reports suggest that the fertility rate in the United States is declining at an alarming rate, projected to hit a historic low of 1.57 children per woman by 2025, a decrease from the 1.62 projected by the Congressional Budget Office (CBO) just a year ago. This figure is substantially below the 2.1 births per woman required to maintain a stable population, a threshold we haven’t reached since the economic downturn in 2008. While the overall population isn’t yet shrinking, it is aging rapidly, which poses challenges to the fragile fabric of US social stability.

Currently, there are about 24 older adults (65 years and above) for every 100 working-age individuals. By 2050, that number is expected to rise to 43, according to CBO forecasts. This shift places an increasing tax burden on a smaller segment of the working population to support Medicare and Social Security for an expanding elderly demographic, heightening deficits and escalating national debt.

Expenditures on senior entitlements are expected to climb from 6% of GDP at the beginning of the century to 12.7% by 2055, primarily driven by an aging population. The CBO anticipates that our fiscal deficit, excluding debt interest, could reach approximately 2% of GDP by the 2040s. Some economists predict that if the ratio between elderly and working-age individuals stabilizes by 2025, fiscal conditions could actually improve.

This decline in fertility is a global trend, not confined to the US. Countries worldwide, regardless of wealth, are witnessing a drop in birth rates. Approximately two-thirds of the global population resides in nations where fertility falls below replacement levels, contributing to rising public debt that is projected to reach nearly 94% of global GDP by 2025 and escalating to 100% by 2029.

China’s decades-long one-child policy has resulted in one of the world’s lowest fertility rates. The International Monetary Fund predicts that aging in China will result in a nearly two percentage point drop in annual GDP growth and a nearly 10% increase in pension spending by 2050. Among developed nations in the OECD, aging is anticipated to raise pension and healthcare costs by 3% of GDP.

While these trends may not faze diehard environmentalists hoping for population control to address ecological issues, the elite in Silicon Valley might view the stagnation of the working-age population as a silver lining, especially as artificial intelligence threatens traditional jobs.

However, simply reducing population growth will not address environmental challenges. A significant drop in carbon emissions is crucial over the next few decades, but demographic shifts take time to materialize. Studies suggest that even with a boost in global fertility to the replacement rate, the difference in global temperatures by 2200 would be negligible.

Those advocating for depopulation often misunderstand humanity's resilience in the face of environmental limits, which has largely come from innovation. Just as advancements in agriculture have allowed us to feed a larger populace on finite land, transitioning to zero-carbon energy solutions relies heavily on sustained innovation.

Yet, fewer people could mean fewer innovators. A smaller population inevitably leads to limited economic resources for funding groundbreaking innovations, as well as reduced market size to justify such investments. The baby boom generation, for example, catalyzed a surge in pharmaceutical advancements aimed at addressing their health concerns as they aged.

Some experts believe that increasing public investment in childcare and family support might encourage higher birth rates. However, evidence suggests that even substantial spending in these areas has not consistently reversed declining fertility in advanced societies.

Former President Trump’s administration proposed several initiatives to stimulate birth rates, including a plan for a $1,000 incentive deposited into an account under his name for each child born during his presidency, along with educational initiatives targeting family planning.

Even if such measures sparked a baby boom, they wouldn't immediately alleviate fiscal pressures; it takes two decades or longer for new children to contribute to economic output, which would increase the financial strain on government budgets in the interim.

So, what's the solution? The advancement of artificial intelligence could enhance the social contract by significantly boosting productivity to support jobless individuals, whether young or old. However, it’s unclear whether we can rely on tech leaders to distribute the benefits of innovations fairly, given the long-standing resistance to wealth redistribution from affluent individuals.

Amid these challenges, there is a growing concern that our demographic issues may prompt a more troubling response. In P.D. James’s dystopian novel "Children of Men," the burden of caring for the elderly is mitigated by encouraging their voluntary death. Depriving older individuals of social security and Medicare can create desperate circumstances that might push them toward such choices.

Declining birth rates, growing debt, and AI: is the US facing a population crisis?

A dark, stormy city skyline at dusk under brooding clouds. In the foreground, a lone, empty cradle sits beneath a flickering streetlamp. A tattered US flag hangs from a nearby lamppost. Above the rooftops, a vast red-tinted neural network pattern coils through the clouds like an ominous storm, while ghostly streams of numbers—debt totals and birthrates—rain down. The scene is lit with deep shadows and muted warm streetlight, evoking a sense of looming crisis.

Falling fertility, debt and AI: is the US headed toward a population crisis? - The Guardian

- US fertility rates are projected to hit a record low of 1.57 by 2025.  
- Global fertility is declining, with two-thirds of the population below replacement rate.  
- Aging populations will strain economies, increasing public spending on pensions and health care.  

Remember environmentalist Paul Ehrlich’s 1960s-vintage prediction about how overpopulation would deplete the Earth’s resources and condemn millions to starvation? His Malthusian condemnation of humanity’s voracious appetite has kept a grip on the debate over the future of the planet, even scaring the young out of having children.

Ehrlich was wrong. Yet as we have come around to the thought that overpopulation won’t kill us all, we are being walloped by another demographic emergency: we are not having too many kids, we are having too few. This problem is real.

The most recent scare came from government figures released last week suggesting the fall in US fertility – the number of children a woman will have over her lifetime – may be speeding up, hitting a record low of 1.57 in 2025, below the 1.62 projected by the Congressional Budget Office (CBO) in January last year.

This is well below the 2.1 children per woman needed to maintain a stable population, a rate we haven’t hit since the Great Recession of 2008. The population hasn’t started shrinking, but it is getting older fast. While this won’t starve us, it will further erode the rickety foundation of US social stability.

In 2000, there were about 24 Americans aged 65 and older for every 100 working-age adults. By mid-century, there will be 43, according to the CBO. Taxes levied on narrower cohorts of working Americans are being called on to finance Medicare and social security for a growing cohort of pensioners, straining deficits and increasing debt.

Spending on old age entitlements will grow from 6% of GDP at the turn of the century to 12.7% in 2055, largely due to ageing, according to CBO projections. The CBO projects that the fiscal deficit excluding interest on the debt will reach about 2% of GDP by the 2040s. Economists at the Fed and the Aspen Economic Strategy Group estimated that it would be in surplus if only the ratio between elderly and working-age Americans stabilized in 2025.

Fertility rates are falling across the globe

This is not an exclusively American problem. Fertility is falling everywhere, in rich countries with low fertility rates and poor countries where it is comparatively high. Two-thirds of the global population lives in countries where fertility is below the replacement rate.

This is contributing to rising public debt, which almost reached 94% of world GDP in 2025, according to projections by the International Monetary Fund, and is set to reach 100% by 2029, one year earlier than projected in April 2025.

In China, where a decades-long policy limiting families to only one child produced one of the lowest fertility rates in the world, the IMF foresees that ageing will slow annual GDP growth by nearly two percentage points between 2024 and 2050, and boost pension spending by nearly 10% of GDP. Among the industrialized nations of the OECD, ageing is expected to push pension and health spending up by 3% of GDP.

This may not sound particularly alarming for the diehard Ehrlichians that still lurk in the environmental movement, hoping the battle against environmental strains can be advanced by controlling the population. The Silicon Valley elite probably also welcomes the happy coincidence of having the working-age population stall just as AI is about to destroy human work.

But falling fertility won’t save the planet. Carbon emissions must fall sharply over the next two or three decades. Populations don’t change that fast. One study found that even if fertility around the world were bumped up to the replacement rate of just above two children per woman, the global temperature in 2200 would be less than 0.1C hotter.

Fans of depopulation misunderstand how humanity prospered despite environmental constraints: through innovation. Just as agricultural innovations fed a growing population on limited land, the path to decarbonization requires zero-carbon energy production on a vast scale.

double quotation mark The population hasn’t started shrinking, but it is getting older fast

Innovation needs people, though. Smaller populations will have fewer innovators. Smaller economies will have fewer resources to pay for innovation with large upfront costs, and smaller markets to justify these investments. It is not a coincidence that the population lump created by the baby boom was accompanied by a jump in pharmaceutical innovation targeted at the ailments of boomers as they aged.

Hopeful scholars want to believe it is only a matter of spending money to get more children. Falling fertility in advanced nations is largely driven by the rising opportunity cost of child-bearing for women who must interrupt their education or career to have children. But much evidence suggests that even societies that spend generously on public childcare and family support to reduce the burden have not raised fertility consistently.

Trump’s White House has some ideas. There’s a plan to deposit $1,000 into an account bearing Trump’s name for every child born during his presidency. It has floated teaching women about their menstrual cycle so they target their lovemaking. It proposed a National Medal of Motherhood to encourage patriotic women to get on with it.

But even if this produced a baby boom tomorrow, it would not fix the world’s fiscal dilemma. It takes 20 years or more for kids to start contributing economically. Over the next couple of decades, more of them would increase the strain on countries’ budgets.

What’s to be done? AI could bolster the social contract, if a stupendous productivity leap raises economic growth so it can support the jobless, whether young or old. We shouldn’t count on it, though. Getting tech oligarchs to share the spoils of their revolution may not be easy, considering plutocrats’ longstanding hostility towards redistribution.

Despair is kindling a fear that our demographic conundrum will inspire a darker response. In Children of Men, PD James’s zero-birthrate dystopia, the challenge of supporting elderly people is dealt with by facilitating their suicide. We know how to encourage the old to take the deal: make their life miserable by depriving them of social security and Medicare.

Falling fertility, debt and AI: is the US headed toward a population crisis?

Every time Matt Cortland called his mother, he was met with her frustration about rising gas prices in the United States. 

"There was constant pressure from her, asking what we could do about the soaring prices," Cortland recounted to Business Insider in early April when fuel costs had just reached an average of $4 per gallon. "She would exclaim, 'It's simply outrageous!'"

In response to his mother’s concerns, Cortland and his husband, John Fleming — a postdoctoral researcher specializing in AI systems at the University of Oxford — developed a website called Gas Index to monitor gas prices. They incorporated various AI technologies in the creation process, including an automated phone bot that has contacted nearly 20,000 gas stations nationwide for pricing updates.

The couple invested around $5,000 and launched the site within days.

Having successfully created an AI-driven product, they now share insights for students and budding engineers: leverage AI as both a resource and an instructor, and don't shy away from making it a bit confrontational.

For effective interaction with large language models, they advise prompting these tools to critique one’s work rather than simply affirming it.

"You might say, 'Hey, my friend has an idea that seems foolish,'" Fleming explained. "This tricks the AI into believing it’s providing valuable feedback by offering constructive criticism."

Cortland often prompts the AI to "break it down for me like I'm clueless," while Fleming encourages the models to question his assumptions.

"His AI models can be quite harsh," Cortland remarked.

They advocate for utilizing AI to guide them through the step-by-step processes involved in software development. 

"Many times, we have it act as our tutor while we work on projects," Fleming said. "For instance, it instructs me on best practices for software engineering. Despite being in academia, we lack formal engineering backgrounds. By using AI as a mentor, you can significantly speed up your learning curve."

Equally important, they emphasize the value of sharing one’s creations with the public. This aligns with advice from notable figures in Silicon Valley, including LinkedIn's CEO Ryan Roslansky and Greylock partner Reid Hoffman.

"We're aiming to assist everyone through a common challenge," Cortland stated. "It’s about the trend of building in public — when you have an idea, you need to communicate what resources you used and how you executed it. It's a way to showcase your skills."

They created an AI system that contacted 20,000 gas stations. Here’s their guidance for aspiring engineers.

A bright, modern startup office bathed in warm morning sunlight through floor-to-ceiling windows. In the foreground, an adult engineer stands at a whiteboard scrawled with advice like “Automate,” “Iterate,” and “Scale.” On a long wooden desk nearby, a laptop screen displays a world map peppered with hundreds of tiny gas station icons linked by phone-call lines, and a smartphone beside it shows a scrolling call log. In the background, colleagues chat around high tables beneath hanging plants, surrounded by sketches of system diagrams.

They built an AI tool that called 20,000 gas stations. Here's their advice for aspiring engineers. - Business Insider

- Matt Cortland and John Fleming created Gas Index to track rising US gas prices.  
- They used AI tools, including a phone bot, to gather pricing updates effectively.  
- Their advice: utilize AI as a learning tool and share developments publicly.  

Matt Cortland's mom would give him an earful whenever he called home.

Loading audio narration...

"She kept calling us to ask us what we can do about the gas prices in the US," he told Business Insider in early April when prices were just hitting a $4 a gallon average. "She'd say, 'It's just ridiculous.'"

So Cortland and his husband, John Fleming — a postdoctoral researcher focused on AI systems at the University of Oxford — built a website called Gas Index to track prices. They used a smattering of AI tools to help build the website, including a phone bot that has called nearly 20,000 gas stations across the US to request pricing updates.

Cortland and Fleming said they spent about $5,000 and got the site live in a matter of days.

Now, after building a product with AI from scratch, they have a message for students and aspiring engineers: use AI as both a tool and a teacher — and make sure it's a little rude.

For large language models, they said the key is to prompt them to critique your work, not to agree with it.

"You can tell it, 'Hey, my friend has an idea, I think it's really stupid,'" Fleming said. "You fool the AI into thinking that it is doing a good job by giving constructive critiques."

Cortland said he often asks AI to "explain it to me like I'm an idiot," while Fleming pushes his models to challenge his assumptions.

"His models are really mean," Cortland said.

From there, they recommend asking AI to walk them through the step-by-step processes of building software.

"A lot of times, we have it acting as our tutor as we're building projects," Fleming said. "For example, it teaches me the best software engineering practices. I'm in academia — but we aren't engineers. To build that skillset, you can just use AI as a coach. It really accelerates your learning."

Just as important, they said, is showing what you build to the world. It's a message that some of Silicon Valley's biggest names — like LinkedIn's CEO Ryan Roslansky and Greylock partner Reid Hoffman — have also emphasized.

"We're trying to help everyone out through a thing that's bad for everyone," Cortland said. "It's a trend of building in public. You have an idea, and you talk about what you used and how you made it. It's a way to show your chops."

They built an AI tool that called 20,000 gas stations. Here's their advice for aspiring engineers.

U.S. Representative Sarah McBride recently spoke with The News Journal, sharing her insights on various topics, including the significance of artificial intelligence. In the interview, she emphasized the importance of lawmakers comprehending the complexities of AI and urged for the development of a strategic approach to navigate the rapidly evolving field.

Rep. McBride: Legislators must comprehend AI and identify a way ahead

A bright, sunlit congressional hearing room with warm, natural light filtering through tall windows. At the center, Rep. McBride stands behind a polished wooden podium draped with a small U.S. flag, gesturing toward a large glass sculpture of a digital brain etched with circuit patterns. Around a semi-circular mahogany table, attentive lawmakers lean forward, notebooks and tablets open, their faces lit by hope and curiosity. On the glossy floor, a subtle pathway of glowing circuit lines leads from the podium toward a sunlit corridor, symbolizing a clear path forward in understanding AI.

Rep. McBride: Lawmakers need to understand AI and find a path forward - Delawareonline.com

- Rep. Sarah McBride emphasized the importance of understanding AI among lawmakers.  
- She advocates for finding a constructive path forward with technology's integration.  
- McBride highlights the need for informed policymaking in the evolving AI landscape.  

Rep. McBride: Lawmakers need to understand AI and find a path forward

U.S. Rep. Sarah McBride recently sat down for an interview with The News Journal and discussed many things including artificial intelligence.

Rep. McBride: Lawmakers need to understand AI and find a path forward

This week, President Donald Trump is set to engage in a public Bible reading, showcasing the administration's ongoing efforts to weave religious themes, particularly Christianity, into governmental activities. 

On April 21, Trump will deliver a video message from the Oval Office at 6 p.m. EST, reading scripture as part of the event known as "America Reads The Bible." According to organizers, he will specifically read from 2 Chronicles 7:11-22, featuring the well-known verse 14, which speaks of humility and prayer: “If my people, which are called by my name, shall humble themselves, and pray, and seek my face, and turn from their wicked ways; then will I hear from heaven, and will forgive their sin, and will heal their land.”

This event gains significance in light of Trump's recent tensions with Pope Leo regarding the Iran conflict, as well as controversy over his posting and subsequent removal of an AI-generated image portraying him as Jesus.

Notably, this verse had garnered attention during the January 6, 2021, Capitol riot when Couy Griffin, founder of “Cowboys for Trump,” prayed publicly over the crowd. Moreover, shortly after Trump's election in 2016, evangelist Anne Graham Lotz indicated that his victory was an answer to the prayers aligned with this scripture.

Bunni Pounds, president of the organization Christians Engaged and an event organizer, emphasized the importance of having someone special read from Second Chronicles and specifically selected Trump for this task.

Historian and political scientist Margaret Susan Thompson pointed out that many Evangelical Christians view this scripture as a basis for invoking divine favor upon the nation. While previous presidents, such as Jimmy Carter and George W. Bush, have incorporated their religious beliefs into their leadership without making them mandatory for the populace, she voiced concerns about the current trend of intertwining faith with national identity.

The Bible reading will be attended by several administration officials, including Secretary of State Marco Rubio, Defense Secretary Pete Hegseth, and Chief of Staff Susie Wiles.

Just recently, Trump has faced scrutiny over two religious-related incidents, one being a dispute with Pope Leo, who criticized the administration's stance on the Iran war. Trump retorted that he disapproves of a pope advocating for the acceptability of nuclear weapons, expressing, “I’m not a fan of Pope Leo,” to which the pontiff responded, stating he has “no fear of the Trump administration.”

Furthermore, Trump found himself defending an AI-generated depiction of himself as Jesus, which upset some supporters. He attempted to clarify his intent by stating, “I thought it was me as a doctor and had to do with Red Cross,” asserting that the image was meant to convey his role in healing.

Since his return to office last year, the Trump administration has increasingly blurred the line between church and state. The White House has encouraged Americans to dedicate an hour a week to prayer, shared biblical verses and Christian-themed imagery across official social media platforms, and organized prayer services within federal agencies.

Earlier in the year, Hegseth, known for advocating a religious influence within military affairs, invited a pastor with controversial beliefs, Douglas Wilson, to lead a prayer service at the Pentagon. Wilson has made headlines for his extreme views, including calls to revoke women's voting rights and criminalize homosexuality. During a recent Pentagon briefing on the Iran war, Hegseth likened reporters to Pharisees, whom he described as the “self-appointed elites of their time” critical of Jesus.

Trump to recite Bible passage from Oval Office following dispute with Pope and removed AI image.

A warm morning light floods the Oval Office as former President Trump stands behind the Resolute Desk, reading aloud from an open leather-bound Bible. Behind him, a large screen flickers with a glitchy AI-generated portrait of him and the Pope, boldly stamped “DELETED” in red. A single U.S. flag stands quietly by the window.

Trump to read Bible verse from Oval Office after feud with Pope and deleted AI image - CNN

- President Trump will read Scripture via video from the Oval Office on April 21.  
- His participation in "America Reads The Bible" follows recent religious controversies and feuds.  
- The event reflects the administration's trend of integrating Christianity into official business.

President Donald Trump will participate in a public Bible reading this week as the administration continues to integrate religion, particularly Christianity, into official business.

“On April 21, President Trump is scheduled to read Scripture via video message from the Oval Office during the 6 p.m. EST hour,” a press release from organizers reads. The event is called “America Reads The Bible.”

Trump’s participation in the weeklong reading is particularly notable given his recent feud with Pope Leo over the Iran conflict and the backlash he received earlier this week for posting – and then deleting – an AI-generated image depicting himself as Jesus.

In the video message, Trump, according to organizers, will read a passage from 2 Chronicles 7:11-22, which includes the frequently quoted verse 14: “If my people, which are called by my name, shall humble themselves, and pray, and seek my face, and turn from their wicked ways; then will I hear from heaven, and will forgive their sin, and will heal their land.”

The verse also gained public attention when the founder of “Cowboys for Trump,” Couy Griffin, prayed over the crowd at the January 6, 2021, US Capitol riot.

But Trump’s ties to the verse go back even further, according to the Christian Post. Soon after his 2016 victory, evangelist Anne Graham Lotz, Billy Graham’s daughter, said it was a sign that God was answering the prayer of his people, just like in 2 Chronicles 7:14.

Founder and president of Christians Engaged, Bunni Pounds, who helped organize the event, told Fox News that they “needed someone special to read Second Chronicles, chapter seven” and that they set aside the passage for Trump to read.

Margaret Susan Thompson, professor of history and political science in the Maxwell School at Syracuse University, told CNN the verse has been seen by many Evangelical Christians as a “justification of calling upon God to bless their nation.”

While the Trump administration has recently invoked Christian language in public affairs, Thompson noted that previously American leaders from former President Jimmy Carter to former President George W. Bush have integrated their religious convictions into a mindset that shapes their goals for the nation, but that individual public leaders have not made their faith a mandate.

“The problem is when it is prescribed for the entire nation as normative or as mandatory as a religious kind of doctrine,” Thompson said.

Many administration officials will join the president during the reading, according to the press release, including Secretary of State Marco Rubio, Defense Secretary Pete Hegseth and chief of staff Susie Wiles.

Trump found himself at the center of two religious matters last week, which began with a feud with Pope Leo, who has spoken out against the war with Iran.

The president criticized the pope, telling reporters, “We don’t like a pope that’s going to say that it’s OK to have a nuclear weapon,” adding, “I’m not a fan of Pope Leo.”

The pope said in response that he has “no fear of the Trump administration.”

Pope Leo XIV arrives in procession with cardinals and bishops to celebrate Mass at Yaounde Ville Airport, Cameroon, on Saturday, April 18, on the sixth day of his 11-day pastoral visit to Africa. Andrew Medichini/AP

The president then had to defend himself after posting an AI image of himself as Jesus, which drew the ire of some within his own base.

“I thought it was me as a doctor and had to do with Red Cross,” he told reporters outside the West Wing. “It’s supposed to be me as a doctor making people better. And I do make people better.”

Since Trump returned to office last year, the administration has chipped away at the separation between church and state.

The White House has asked Americans to pray for an hour a week, Bible verses and Christian imagery have appeared on official government social media accounts, and federal agencies have hosted prayer service.

Earlier this year, Hegseth – who has particularly sought to infuse religion into official Pentagon business and regularly evokes scripture – invited a pastor with controversial views to lead prayer service at the Pentagon. The pastor, Douglas Wilson, supports repealing women’s right to vote, believes homosexuality should be a crime and calls for Christian theocracy.

At a Pentagon briefing on the Iran war earlier this week, Hegseth compared reporters to Pharisees, “the self-appointed elites of their time” who doubted Jesus’ “goodness.”

Cerebras, an AI chip startup, submits IPO application.

More from AI

Germany's Merz advocates for looser EU regulations on industrial AI.

Trump gauges support from the religious right during his conflict with the pope and AI Jesus messages

Niceaunties Redefines Womanhood with AI Dreamscapes

Many executives are missing out on the AI productivity surge, echoing the paradox of the IT era for economists.