Nvidia CEO Jensen Huang says the world is “at the beginning of the AI revolution,” and the rapid adoption of artificial intelligence across industries could bring “probably” a transition to four-day work weeks, marking another shift in social behavior akin to previous industrial revolutions. But that doesn’t mean life will slow down.
“I have to admit that I’m afraid to say that we are going to be busier in the future than now,” Huang told Liz Claman on Fox Business Network’s The Claman Countdown. He pointed to AI’s uncanny ability to take time-consuming things and get them done very quickly, and predicted that the way this will actually work is to realize the ideas of more business leaders with many ideas in their heads, like himself. “I’m always waiting for work to get done because I’ve got more ideas,” he told Claman, adding that he thinks “most companies have more ideas than we know what to pursue. And so the more productive we are, the more opportunity we get to go pursue new ideas.”
“Every industrial revolution leads to some change in social behavior,” Huang noted in a wide-ranging interview, predicting that GDP will grow and productivity will increase. Certainly his company is growing. Huang was speaking to Claman in the aftermath of Nvidia’s record $46.7 billion second quarter earnings announced just a few days beforehand. His company still has a market capitalization north of $4 trillion, the largest in the world.
Bank of America Research has predicted a sweeping productivity boom for the S&P 500 as companies learn to do more with less, solving the “productivity paradox” that has been a characteristic of much of the computer age: you can see the revolution everywhere except in the productivity statistics. BofA Research’s Head of US Equity & Quantitative Strategy, Savita Subramanian, told Fortune this was partially, but not entirely, due to AI. The key is the same thing identified by Huang: doing things more efficiently. “If you’re productive, you are doing things more efficiently, you need less labor. And this is more labor efficiency than anything else.”
Extraordinary growth and demand for AI
Huang emphasized that AI capabilities now touch nearly every sector, from cloud computing to manufacturing, robotics, and even self-driving vehicles. He detailed the explosive growth in demand for Nvidia’s AI chips, especially their new Blackwell Ultra architecture (code-named GB300), fueled by surging global investments in data centers. Huang estimated that through the end of the decade, about $3 to $4 trillion of AI factory infrastructure will be built out.
Looking to the future, Huang acknowledged both anxiety and excitement about AI’s effect on work.
The interview also touched on the geopolitical tension of chip exports to China and the Trump administration’s stance on licensing. Huang positioned US technology as a potential global standard, remarking, “Having the world build AI on American tech stack helps America win.” Nvidia remains eager to resume China shipments, which could reclaim a share of a $50 billion AI hardware market there.
The 4-day week is already reality in some places
Fortune highlights that several companies pioneering the four-day work week have seen notable wins: productivity climbed by up to 24%, burnout was halved, and turnover dropped sharply. Large-scale studies in Britain and North America found that workers can accomplish the same results in around 33 to 34 hours weekly—and the drop from five to four days led to significantly better health, job satisfaction, and a dramatic reduction in sick days and quitting rates, suggesting that the current five-day work week is largely performative.
In the Netherlands, Fortune has reported, workers routinely put in just 32 hours a week, enjoying the quality-of-life improvements that come from a four-day schedule. Employees overwhelmingly want to keep the shorter week after pilot programs end and organizations that switch to them rarely revert to five days, supporting Huang’s assertion that industrial revolutions bring lasting social change.
Huang noted that with the advent of modern capitalism, the seven- or six-day work week evolved into a a world of five-day work weeks. “Every industrial revolution leads to some change in social behavior,” he said, adding that he expects the economy to be doing very well because of AI and automation.” He returned to the hotly debated impact on jobs, where Huang has been optimistic and peers such as Anthropic’s Dario Amodei have predicted that 50% of white-collar work will vanish. “Some jobs will go away,” Huang said. “Many jobs will be new and invented. But one thing for sure, every job will be changed as a result of AI.” He added that he believes “life quality will get better, of course, over time.”
Nvidia declined to comment further.
For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.
A case that could provide the Trump administration with new and expansive power over the traditionally independent Federal Reserve focused Friday on what would constitute “cause” to remove a high-ranking official at the nation’s central bank.
Federal Reserve Governor Lisa Cook has requested an emergency injunction—in the form of a temporary restraining order—to block President Donald Trump’s attempt to fire her over allegations that she committed mortgage fraud when she purchased a home and condo in 2021. She was appointed to the Fed’s board by former president Joe Biden in 2022.
Arguments in court Friday centered on what constitutes “cause,” which in this case are unproven accusations by a Trump appointee that Cook committed mortgage fraud.
In an exchange with U.S. District Judge Jia Cobb, Cook’s lawyer, Abbe David Lowell, said Trump’s motivations are clear. “He’s already said he wants a majority (on the Fed board). He’s bragged that he’s going to get it.’’
If her firing is allowed to stand, it would likely erode the Fed’s longstanding independence from day-to-day politics. No president has ever fired a Fed governor in the agency’s 112-year history. Economists broadly support Fed independence because it makes it easier for the central bank to take unpopular steps such as raising interest rates to combat inflation.
Cook has asked the court to issue an emergency order that would prevent her firing and enable Cook to remain on the seven-member board of governors while her lawsuit seeking to overturn the firing makes its way through the courts. Many observers expect her case will end up at the U.S. Supreme Court.
In court Friday, the Justice Department’s Yaakov Roth, who represented the Trump administration at the hearing, complained that Cook had not offered an explanation for anything questionable in her mortgage documents or a defense against the fraud allegations.
The allegations remain just that, leveled by Bill Pulte, Trump’s appointee to the agency that oversees mortgage giants Fannie Mae and Freddie Mac.
And late Thursday, Pulte, said that Cook had allegedly committed fraud on a third property, a condominium in Cambridge, Massachusetts, in April 2021.
Pulte said in a social media post that Cook classified the condo as a “second home,” but in financial disclosure documents filed in 2022 through 2025 with the government, she described it as an “investment/rental” property. Pulte added that mortgage rates and down payments for second homes can be lower than for investment properties.
Pulte also alleged, without evidence, that Cook may have rented out two properties in Ann Arbor, Mich. and Atlanta, which were the focus of a criminal referral he made last week and which he said she has claimed as her principal residence.
In a statement, Cook’s lawyer, Lowell, decried “an obvious smear campaign aimed at discrediting Gov. Cook … Nothing in these vague, unsubstantiated allegations has any relevance to Gov Cook’s role at the Federal Reserve, and they in no way justify her removal from the Board.”
The law governing the Fed says the president can’t fire a governor just because they disagree over interest rate policy. Trump has repeatedly demanded that the Fed, led by Chair Jerome Powell, reduce its key interest rate, which is currently 4.3%. Yet the Fed has kept it unchanged for the last five meetings.
But the president may be able to fire a Fed governor “for cause,” which has traditionally been interpreted to mean inefficiency, neglect of duty, or malfeasance. Cook’s lawyers argue that it also refers only to conduct while in office. They also say that she was entitled to a hearing and an opportunity to rebut the charges.
“The unsubstantiated and unproven allegation that Governor Cook ‘potentially’ erred in filling out a mortgage form prior to her Senate confirmation — does not amount to ‘cause,’” the lawsuit says.
The Supreme Court declined to temporarily block the president from firing directors of some independent agencies earlier this year while those cases move through the courts. Legal experts say the high court this year has shown more deference to the president’s removal powers than it has in the past.
Still, in a case in May, the Supreme Court appeared to single out the Fed as deserving of greater independence than other agencies, describing it as “a uniquely structured, quasi-private entity.” As a result, it’s harder to gauge how the Supreme Court could rule if this case lands in its lap.
As a governor, Cook votes on all the Fed’s interest rate decisions and helps oversee bank regulation. The Fed has substantial power over the economy by raising or cutting its key interest rate, which can then influence a broad range of other borrowing costs, including mortgages, car loans, and business loans.
Pulte’s charge that Cook has committed mortgage fraud is one he has also made against two of Trump’s biggest political enemies, California Democratic Sen. Adam Schiff and New York Attorney General Letitia James, who has prosecuted Trump. Pulte has ignored a similar case involving Ken Paxton, the Texas attorney general who is friendly with Trump and is running for Senate in his state’s Republican primary.
Cook’s lawsuit responds by arguing that the claims are just a pretext “in order to effectuate her prompt removal and vacate a seat for President Trump to fill and forward his agenda to undermine the independence of the Federal Reserve.”
If Trump can replace Cook, he may be able to gain a 4-3 majority on the Fed’s governing board. Trump appointed two board members during his first term and has nominated a key White House economic adviser, Stephen Miran, to replace Adriana Kugler, another Fed governor who stepped down unexpectedly Aug. 1. Trump has said he will only appoint people to the Fed who will support lower rates.
Federal Reserve Chairman Jerome Powell, left, talks with Board of Governors member Lisa Cook, right, during an open meeting of the Board of Governors at the Federal Reserve, June 25, 2025, in Washington.
A new survey reveals a striking “AI readiness gap” in the modern workplace: Those using AI tools the most—including top executives and Gen Z employees—are often the least likely to receive meaningful guidance, training, or even company approval for their use.
The findings come from WalkMe, an SAP company, which surveyed over 1,000 U.S. workers for the 2025 edition of its AI in the Workplace survey. Nearly half of employees (48.8%) admit to hiding their use of AI at work to avoid judgment, suggesting that something like “AI shame” is a real phenomenon in the workplace. This discomfort is especially pronounced at the top, with 53.4% of C-suite leaders admitting they conceal their AI habits—despite being the most frequent users. Entry-level workers aren’t exempt, but the paradox deepens at the executive level, highlighting how even the most empowered employees remain uneasy.
Gen Z: Eager, but unsupported
Gen Z approaches AI with both enthusiasm and anxiousness. A striking 62.6% have completed work using AI but pretended it was all their own effort—the highest rate among any generation.
More than half (55.4%) have feigned understanding of AI in meetings. Their behavior is context-dependent: 28.4% exaggerate their AI use to some, while 13.5% downplay it to others; it depends on whom they’re speaking with. But only 6.8% report receiving extensive, time-consuming AI training, and 13.5% received none at all. This is the lowest of any age group. Despite this, an overwhelming 89.2% use AI at work—and just as many (89.2%) use tools that weren’t provided or sanctioned by their employer. Only 7.5% reported receiving extensive training with AI tools. This is a strikingly small advance from 2024, when the same survey from WalkMe found 7.0% reported extensive training—just a 0.5% increase.
Sharon Bernstein, chief human resources officer for WalkMe, told Fortune in an interview, “Companies are not educating enough about this whole thing,” adding that they seem to not be facilitating use of AI tools. They “are not training their employees enough today, or guiding … Even if you are an amazing CIO, and you’re allowed to buy a few different tools for AI, how much was it adopted? Like, for real?”
The AI class divide and a productivity paradox
Access to AI training and guidance increases with rank and company size. Only 3.7% of entry-level employees receive substantial training compared with 17.1% of C-level executives. Younger and junior staff remain unsupported—a gap that risks cementing an “AI class divide” where the most frequent users are left to navigate on their own.
AI is changing work, and the survey suggests not always for the better. Most employees (80%) say AI has improved their productivity, but 59% confess to spending more time wrestling with AI tools than if they’d just done the work themselves. Gen Z again leads the struggle, with 65.3% saying AI slows them down (the highest amount of any group), and 68% feeling pressure to produce more work because of it. Nearly one in three are deeply anxious about AI’s impact on their jobs, saying they worry “a lot” about this issue. Confidence is mixed: Only 45% of Gen Zers say they’re “very confident” using AI—less than millennials (56.3%), and nearly equal with Gen X (43.2%).
How this fits into the picture
These gaps around AI readiness and varying levels of AI shame fit into an emerging picture of a confusing, if not chaotic, implementation of AI in the workplace, from the entry level all the way to the C-suite. For instance, more than half of professionals report being overwhelmed by AI training initiatives, saying that it feels like “a second job”—adding stress and longer hours, often with little tangible benefit to workflows. While it’s speculative to link lack of proper training to the bombshell MIT study showing a staggering 95% failure rate for generative AI pilots at large enterprises, there is clearly an issue going from the drawing board to the factory floor. Furthermore, this disconnection between corporate hype and actual business value is fueling investor worries about a potential AI bubble.
Another major study, the first of its kind in the field, came out from Stanford and top economist Erik Brynjolfsson, a thought leader in AI. His team found that since late 2022, when generative AI exploded onto the scene, there has been a statistically significant decline in entry-level hiring, in jobs directly exposed to AI automation. This means that mastery of AI tools will be hugely important for entry-level workers, and the WalkMe survey suggests they are getting the least amount of training.
Finally, the survey confirms the trend in “shadow AI,” where workers are overwhelmingly using these tools, but companies are further behind in their official adoption. Many colleges are banning AI tools, meanwhile, as they try to stem what they perceive as a rampant “cheating” crisis. From the market, where investors fear a bubble; to the entry level, where workers are trying to match their shadow use of AI to their actual performance; to the C-suite, where leaders are under pressure to revolutionize their companies and get results with this new technology, there’s an emerging gap between theory and reality.
Bernstein spoke from her perspective as an HR leader: “First of all, you want people not to fear to admit that they use it, right?” She urged companies to be transparent about how they are planning to use AI, to displace the fear that AI tools will replace workers, on the one hand, and to promote facility with using AI, on the other hand. “I don’t really think that we can literally replace employees,” she added. “Maybe in very specific positions, but in general, I think companies are now in a stage that they need to educate their team members about it.”
Rising anxiety, falling readiness
Worry about AI’s effect on jobs is intensifying: 44.8% of workers are worried, and the proportion “very worried” has spiked since last year. Gen Z feels this most acutely: 62.2% say they worry about AI’s impact, with 28.4% “very worried”—the highest rate across age groups. Stress levels are up for 27% of Gen Z, the highest of any generation. Yet hope persists: 89.6% want to learn more about AI, and 86% believe AI proficiency is critical for career success.
The findings point to an urgent need for employers to bridge the AI readiness gap, offering clear guidance, comprehensive training, and transparent policies. Those on the leading edge of AI adoption—whether in the boardroom or among Gen Z—need support, not secrecy. As tools proliferate and expectations rise, organizations risk eroding trust, productivity, and emotional well-being unless this issue is addressed head-on.
For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.
Over half of professionals report that AI trainings feel like a second job, according to a recent LinkedIn survey, highlighting widespread frustration among workers with the proliferation of workplace automation programs.
A majority of respondents (51%) find the intensity and frequency of AI training requirements excessive, stating that it’s interfering with their core job responsibilities and contributing to burnout. Employees cited dense training modules, unrealistic deadlines, and a lack of clarity about practical benefits as key sources of dissatisfaction.
LinkedIn found an 82% increase in people posting on the platform about feeling overwhelmed and navigating change this year. “The mounting pressure to upskill in AI is fueling insecurity among professionals at work—with a third (33%) admitting they feel embarrassed by how little they understand it, and 35% saying they feel nervous talking about AI at work for fear of sounding uninformed,” LinkedIn wrote.
Workplace impact
These findings come as employers increase investment in upskilling efforts designed to help staff adapt to new AI-based processes. Instead of feeling empowered, many professionals say these trainings add stress and extend their working hours, often without extra compensation or real improvements to workflow.
There are real consequences for this and anecdotal evidence that workers are justified in feeling insecure. IgniteTech CEO Eric Vaughan told Fortune earlier this month that he laid off nearly 80% of his staff after they failed to respond to AI training, while Joshua Wöhle of Mindstone relayed a similar story of a client-CEO who ordered his staff to dedicate all Fridays to AI retraining, and invited them to leave the company if they didn’t report back constructively on their findings.
The survey also found that, amid the flood of AI-related content and programs, professionals are increasingly turning to their networks—rather than AI tools or search engines—for trusted advice and support in navigating workplace changes. Some 43% of professionals say “their network, the people they know, is still their No. 1 source for advice at work,” ahead of search engines and AI tools. Nearly two-thirds (64%) of professionals say colleagues are helping them make decisions faster and more confidently.
Mounting frustration with mandatory AI trainings may be just the tip of the iceberg. A recent MIT study found that 95% of generative AI pilots at enterprises have failed to deliver any measurable return on investment—fueling growing concerns over an AI stock bubble as corporate spending and investor hype far outweigh results. It seems to be tied to this frustration over ineffective or stumbling AI training efforts.
MIT’s sobering findings
The MIT NANDA report analyzed hundreds of AI deployments and found only 5% produced rapid revenue acceleration or noticeable operational improvements. The majority of pilots stall in the testing phase or get abandoned, with large companies taking nearly a year to scale projects that rarely succeed. Flawed enterprise integration and a gap in AI literacy—not just model quality—were cited as the main barriers.
Wall Street and institutional investors are sounding the alarm, worried that record AI investments aren’t translating to profits and could trigger a painful reckoning for overvalued tech stocks. Some have started trimming exposure, fearing that the gap between reality and hype may be unsustainable, reminiscent of prior tech bubbles. The all-important Nvidia earnings on Wednesday illustrate the jitters, as record revenue still failed to prevent investors taking a few percentage points off the stock.
Connections to workforce concerns
As companies pour money into AI pilots and tech stocks, employees are increasingly skeptical of both the business value and the constant upskilling requirements. With over half of professionals saying AI trainings feel like a second job, the MIT report adds new context: Companies’ aggressive push for digital transformation is straining workers, not yet augmenting them, as widely billed.
The results underscore mounting tension between the pace of technological implementation and the lived experience of professionals, suggesting that companies may need to rethink their approach to AI upskilling to avoid further alienating employees.
Update, Aug. 28, 2025: This report has been updated to clarify that workers undergoing AI training do not necessarily find it annoying.
For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.
The U.S. housing market’s inventory is growing, putting pressure on prices and slowing new construction, according to fresh research from the Bank of America Institute. As of June, existing-home supply reached 4.7 months, the highest level since July 2016. New-home supply surged even further to 9.8 months—its highest point since 2022—highlighting how quickly inventory is building across the housing market.
The influx of available homes reflects sluggish demand, with builders citing weak buyer urgency, affordability challenges, and lingering job instability. The Institute noted new-home inventory is now at its highest level since 2007, the year before the housing market collapse that led to the Great Financial Crisis.
ResiClub co-founder Lance Lambert told Fortune that the rising inventory tells us that “homebuyers are gaining leverage” as slack in the housing market is increasing. “The Pandemic Housing Boom saw too much housing demand all at once, home prices overheated too fast in many markets, and underlying fundamentals got too stretched.”
Lambert characterized the last few years as a “recalibration period” where the housing market is smoothing out that excess. Mounting inventory sucks out appreciation in more markets—and even causes outright corrections in some markets’ home prices. He said he expects the underlying fundamentals to slowly improve as that happens and incomes keep rising. “It takes time.” This period is different from 2007, he said, because that window saw a far greater weakening of the housing market and upswing in resale inventory, along with unsold, completed newbuild homes.
BofA Research
One striking shift: The median price of a new home has actually fallen below that of an existing home—a reversal of the usual market dynamic. BofA said this pricing inversion underscores how builders are being forced to discount amid rising supply and softer demand. “Builders are starting to pull back on new home starts in many markets,” Bank of America wrote. While the slowdown is broad-based, conditions vary regionally, with some areas such as the Midwest proving more resilient than others.
“Since the Pandemic Housing Boom fizzled out in 2022, and the affordability squeeze was fully felt,” Lambert told Fortune, “the national power dynamic has slowly been shifting from sellers to buyers as homes have a harder time selling and active inventory for sale builds.”
Still, Lambert noted the inventory picture varies significantly across the country. For instance, it remains most limited across notable sections of the Midwest and the Northeast, although still growing, he said. On the other hand, active inventory has neared or surpassed pre-pandemic 2019 levels in many parts of the Sun Belt and Mountain West, and he said that is where homebuyers have gained the most leverage.
The trend comes as the Federal Reserve has begun trimming interest rates in an effort to support both broader economic growth and housing affordability. Whether those cuts will be enough to reignite demand remains an open question.
For now, the data signals a market in transition: high inventory, moderating prices, and builders caught between a cautious consumer and the need to manage supply.
For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.
First it was the release of GPT-5 that OpenAI “totally screwed up,” according to Sam Altman. Then Altman followed that up by saying the B-word at a dinner with reporters. “When bubbles happen, smart people get overexcited about a kernel of truth,” The Verge reported on comments by the OpenAI CEO. Then it was the sweeping MIT survey that put a number on what so many people seem to be feeling: a whopping 95% of generative AI pilots at companies are failing.
A tech sell-off ensued, as rattled investors sent the value of the S&P 500 down by $1 trillion. Given the increasing dominance of that index by tech stocks that have largely transformed into AI stocks, it was a sign of nerves that the AI boom was turning into dotcom bubble 2.0. To be sure, fears about the AI trade aren’t the only factor moving markets, as evidenced by the S&P 500 snapping a five-day losing streak on Friday after Jerome Powell’s quasi-dovish comments at Jackson Hole, Wyoming, as even the hint of openness from the Fed chair toward a September rate cut set markets on a tear.
Gary Marcus has been warning of the limits of large language models (LLMs) since 2019 and warning of a potential bubble and problematic economics since 2023. His words carry a particularly distinctive weight. The cognitive scientist turned longtime AI researcher has been active in the machine learning space since 2015, when he founded Geometric Intelligence. That company was acquired by Uber in 2016, and Marcus left shortly afterward, working at other AI startups while offering vocal criticism of what he sees as dead-ends in the AI space.
Still, Marcus doesn’t see himself as a “Cassandra,” and he’s not trying to be, he told Fortune in an interview. Cassandra, a figure from Greek tragedy, was a character who uttered accurate prophecies but wasn’t believed until it was too late. “I see myself as a realist and as someone who foresaw the problems and was correct about them.”
Marcus attributes the wobble in markets to GPT-5 above all. It’s not a failure, he said, but it’s “underwhelming,” a “disappointment,” and that’s “really woken a lot of people up. You know, GPT-5 was sold, basically, as AGI, and it just isn’t,” he added, referencing artificial general intelligence, a hypothetical AI with human-like reasoning abilities. “It’s not a terrible model, it’s not like it’s bad,” he said, but “it’s not the quantum leap that a lot of people were led to expect.”
Marcus said this shouldn’t be news to anyone paying attention, as he argued in 2022 that “deep learning is hitting a wall.” To be sure, Marcus has been wondering openly on his Substack on when the generative AI bubble will deflate. He told Fortune that “crowd psychology” is definitely taking place, and he thinks every day about the John Maynard Keynes quote: “The market can stay irrational longer than you can stay solvent,” or Looney Tunes’s Wile E. Coyote following Road Runner off the edge of a cliff and hanging in midair, before falling down to Earth.
“That’s what I feel like,” Marcus says. “We are off the cliff. This does not make sense. And we get some signs from the last few days that people are finally noticing.”
Building warning signs
The bubble talk began heating up in July, when Apollo Global Management’s chief economist, Torsten Slok, widely read and influential on Wall Street, issued a striking calculation while falling short of declaring a bubble. “The difference between the IT bubble in the 1990s and the AI bubble today is that the top 10 companies in the S&P 500 today are more overvalued than they were in the 1990s,” he wrote, warning that the forward P/E ratios and staggering market capitalizations of companies such as Nvidia, Microsoft, Apple, and Meta had “become detached from their earnings.”
In the weeks since, the disappointment of GPT-5 was an important development, but not the only one. Another warning sign is the massive amount of spending on data centers to support all the theoretical future demand for AI use. Slok has tackled this subject as well, finding that data center investments’ contribution to GDP growth has been the same as consumer spending over the first half of 2025, which is notable since consumer spending makes up 70% of GDP. (The Wall Street Journal‘s Christopher Mims had offered the calculation weeks earlier.) Finally, on August 19, former Google CEO Eric Schmidt co-authored a widely discussed New York Times op-ed on August 19, arguing that “it is uncertain how soon artificial general intelligence can be achieved.”
This is a significant about-face, according to political scientist Henry Farrell, who argued in the Financial Times in January that Schmidt was a key voice shaping the “New Washington Consensus,” predicated in part on AGI being “right around the corner.” On his Substack, Farrell said Schmidt’s op-ed shows that his prior set of assumptions are “visibly crumbling away,” while caveating that he had been relying on informal conversations with people he knew in the intersection of D.C. foreign policy and tech policy. Farrell’s title for that post: “The twilight of tech unilateralism.” He concluded: “If the AGI bet is a bad one, then much of the rationale for this consensus falls apart. And that is the conclusion that Eric Schmidt seems to be coming to.”
Finally, the vibe is shifting in the summer of 2025 into a mounting AI backlash. Darrell West warned in Brookingsin May that the tide of both public and scientific opinion would soon turn against AI’s masters of the universe. Soon after, Fast Company predicted the summer would be full of “AI slop.” By early August, Axios had identified the slang “clunker” being applied widely to AI mishaps, particularly in customer service gone awry.
History says: short-term pain, long-term gain
John Thornhill of the Financial Times offered some perspective on the bubble question, advising readers to brace themselves for a crash, but to prepare for a future “golden age” of AI nonetheless. He highlights the data center buildout—a staggering $750 billion investment from Big Tech over 2024 and 2025, and part of a global rollout projected to hit $3 trillion by 2029. Thornhill turns to financial historians for some comfort and some perspective. Over and over, it shows that this type of frenzied investment typically triggers bubbles, dramatic crashes, and creative destruction—but that eventually durable value is realized.
He notes that Carlota Perez documented this pattern in Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages. She identified AI as the fifth technological revolution to follow the pattern begun in the late 18th century, as a result of which the modern economy now has railroad infrastructure and personal computers, among other things. Each one had a bubble and a crash at some point. Thornhill didn’t cite him in this particular column, but Edward Chancellor documented similar patterns in his classic Devil Take The Hindmost, a book notable not just for its discussions of bubbles but for predicting the dotcom bubble before it happened.
Owen Lamont of Acadian Asset Management cited Chancellor in November 2024, when he argued that a key bubble moment had been passed: an unusually large number of market participants saying that prices are too high, but insisting that they’re likely to rise further.
Wall Street is cautious, but not calling a bubble
Wall Street banks are largely not calling for a bubble. Morgan Stanley released a note recently seeing huge efficiencies ahead for companies as a result of AI: $920 billion per year for the S&P 500. UBS, for its part, concurred with the caution flagged in the news-making MIT research. It warned investors to expect a period of “capex indigestion” accompanying the data center buildout, but it also maintained that AI adoption is expanding far beyond expectations, citing growing monetization from OpenAI’s ChatGPT, Alphabet’s Gemini, and AI-powered CRM systems.
Bank of America Research wrote a note in early August, before the launch of GPT-5, seeing AI as part of a worker productivity “sea change” that will drive an ongoing “innovation premium” for S&P 500 firms. Head of U.S. Equity Strategy Savita Subramanian essentially argued that the inflation wave of the 2020s taught companies to do more with less, to turn people into processes, and that AI will turbo-charge this. “I don’t think it’s necessarily a bubble in the S&P 500,” she told Fortune in an interview, before adding, “I think there are other areas where it’s becoming a little bit bubble-like.”
Subramanian mentioned smaller companies and potentially private lending as areas “that potentially have re-rated too aggressively.” She’s also concerned about the risk of companies diving into data centers too such a great extent, noting that this represents a shift back toward an asset-heavier approach, instead of the asset-light approach that increasingly distinguishes top performance in the U.S. economy.
“I mean, this is new,” she said. “Tech used to be very asset-light and just spent money on R&D and innovation, and now they’re spending money to build out these data centers,” adding that she sees it as potentially marking the end of their asset-light, high-margin existence and basically transforming them into “very asset-intensive and more manufacturing-like than they used to be.” From her perspective, that warrants a lower multiple in the stock market. When asked if that is tantamount to a bubble, if not a correction, she said “it’s starting to happen in places,” and she agrees with the comparison to the railroad boom.
The math and the ghost in the machine
Gary Marcus also cited the fundamentals of math as a reason that he’s concerned, with nearly 500 AI unicorns being valued at $2.7 trillion. “That just doesn’t make sense relative to how much revenue is coming [in],” he said. Marcus cited OpenAI reporting $1 billion in revenue in July, but still not being profitable. Speculating, he extrapolated that to OpenAI having roughly half the AI market, and offered a rough calculation that it means about $25 billion a year of revenue for the sector, “which is not nothing, but it costs a lot of money to do this, and there’s trillions of dollars [invested].”
So if Marcus is correct, why haven’t people been listening to him for years? He said he’s been warning people about this for years, too, calling it the “gullibility gap” in his 2019 book Rebooting AI and arguing in The New Yorker in 2012 that deep learning was a ladder that wouldn’t reach the moon. For the first 25 years of his career, Marcus trained and practiced as a cognitive scientist, and learned about the “anthropomorphization people do. … [they] look at these machines and make the mistake of attributing to them an intelligence that is not really there, a humanness that is not really there, and they wind up using them as a companion, and they wind up thinking that they’re closer to solving these problems than they actually are.” He said he thinks the bubble inflating to its current extent is in large part because of the human impulse to project ourselves onto things, something a cognitive scientist is trained not to do.
These machines might seem like they’re human, but “they don’t actually work like you,” Marcus said, adding, “this entire market has been based on people not understanding that, imagining that scaling was going to solve all of this, because they don’t really understand the problem. I mean, it’s almost tragic.”
Subramanian, for her part, said she thinks “people love this AI technology because it feels like sorcery. It feels a little magical and mystical … the truth is it hasn’t really changed the world that much yet, but I don’t think it’s something to be dismissed.” She’s also become really taken with it herself. “I’m already using ChatGPT more than my kids are. I mean, it’s kind of interesting to see this. I use ChatGPT for everything now.”
What is a “sea change” and what does it have to do with worker productivity, the computer age, and the “productivity paradox?” The first phrase comes to us from Shakespeare, and it means a sort of mystical transformation, after which something is fundamentally different from before. The second phase comes from Nobel laureate Robert Solow, about how you can see the computer age everywhere but in the productivity statistics. Bank of America Research thinks it sees a way that workers really are getting more productive—and AI is only one part of the puzzle.
The late plays by William Shakespeare are alternately called his “romances” or his “problem plays,” because of their ambiguity in tone, as they alternate from passages of magical realism to stark scenes that grapple with complex social issues. At times, they point the way toward the prestige TV of the early 21st century where, for instance, The Sopranos could range from broad comedy to intense violence to avant-garde dream sequences, all in one episode. It’s from the romances that we get phrases that stick with us today, like the description from The Tempest of a “sea change into something rich and strange.”
Full disclosure: The author’s brother is an eminent Shakespearean scholar, often quoted in The New York Times, although never previously in Fortune, and so I asked him to explain what this particular term means. “Toward the end of his career,” Drew Lichtenberg of the Shakespeare Theatre Company in Washington DC, said in a statement to Fortune, “Shakespeare started writing genre-defying plays with sudden and miraculous changes of fortune.” Shakespeare used the phrase “sea change” to describe a “magical storm at sea that has the power to snuff out life or restore it in less than a second.”
What do Shakespeare’s plays of miraculous changes of fortune have to do with, well, Fortune? Bank of America Institute has projected a “sea change” in the economy. It sees a pivotal transformation in worker productivity at America’s largest companies, driven by lessons from post-pandemic inflation and supercharged by a wave of artificial intelligence and automation. The institute worked hand in hand with projections from Bank of America Research to project a rewiring of the fundamental valuation landscape of the S&P 500, with profound implications for investors and the “quality premium” that U.S. stocks traditionally command.
Fortune talked to BofA Research’s Head of US Equity & Quantitative Strategy, Savita Subramanian, to dig into this change, potentially to something rich and strange. It’s not quite that mystical, she said, but she still thinks it’s a big deal.
Finally, a productivity surge?
Subramanian explained that what her team has projected isn’t as exciting or dramatic as having actual wizards working at the gears of the economy. The more prosaic insight, she says, is that the combination of AI technology and lessons learned from the inflation wave of the 2020s mean that worker productivity is finally showing signs of increasing. That’s the sea change taking place.
At its heart, her work is all about the famous “productivity paradox” identified by Nobel prize-winning economist Robert Solow. “You can see the computer age everywhere but in the productivity statistics,” he said in 1987, long before the productivity crisis of the 21st century set in. As Fortune‘s Jeremy Kahn has discussed, workers still don’t seem to be getting more productive despite the bevy of new technologies at their disposal. In fact, McKinsey’s Chris White and Olivia White argued in 2024 that productivity has been dismal for nearly a generation, hovering around 1% a year, with a dip after the Great Financial Crisis. Subramanian agrees, telling Fortune that if you look at productivity measures, “they haven’t really improved all that much since 2001.”
Subramanian wrote on Aug. 8 that the end goal of the massive AI spending that’s rippling through the economy is a “sea-change” in the scale and scope of efficiency gains—and this productivity cycle is already under way. Post-pandemic wage inflation forced companies “to do more with fewer people,” she added, and now AI tools are due to kick that up a notch.
But the official stats don’t show a complete understanding of how productivity really functions, Subramanian explained. So BofA took sales, adjusted for inflation, and then divided sales by the number of people working at S&P 500 companies, showing real sales growth versus number of people, what she called a “decent proxy” for productivity, “because if you’re productive, you are doing things more efficiently, you need less labor. And this is more labor efficiency than anything else.”
Look at what she found.
This means companies are learning to do more with less, and that is kind of magical. Companies have had to do harder work to generate earnings and keep margins healthy, often by replacing their people with processes. “A process is almost free and it’s replicable for eternity,” she said, adding that she thinks this is why the companies exercising efficiency gains have tended to outperform. It’s not only about AI displacing workers, but a fundamental shift in how business is being done.
‘It feels like sorcery’
This discussion may seem on its face to be more boring than a tempest and a wizard, she said, but there is something supernatural about the current moment. “I think people love this AI technology because it feels like sorcery,” she said, before adding, “the truth is it hasn’t really changed the world that much yet, but I don’t think it’s something to be dismissed.”
Overall, Subramanian finds the S&P 500 has shifted from its 1980s model of asset- and labor-intensive manufacturing to asset- and labor-light innovation, namely tech and health care firms. Showing her work, she calculates that the S&P 500 firms with a focus on innovation, measured through high research and development expenditures, trade at structurally higher multiples of 29x forward earnings per share. More capital-intensive manufacturers, on the other hand, trade at a 21x multiple. The current AI boom is actually a bit risky, she wrote, because the massive data center investments represent a shift from an asset-light to an asset-heavier focus.
To be sure, BofA finds that the S&P 500 is now statistically expensive on 19 out of the 20 metrics that they track, including P/E, price to book, price to cash flow, and market cap/GDP. That’s where the sea change matters, because if the shift from manufacturing to innovation is real, then valuations have to shift as well. Hence the “innovation premium” from BofA’s research.
Excluding Tesla, Subramanian talks about the other members of the “Magnificent Seven” as evidence of firms losing some of their innovation premium as a result of a shift toward asset-heaviness. As a basket of stocks, Microsoft, Google, Amazon, Meta, Nvidia and Apple’s average shareholder yield (i.e., dividends plus net buybacks) has dropped by over 1% since 2015.
There are other shifts afoot as well, she told Fortune. “We seem to be at least pausing on this globalization theme,” she said, citing China’s admission to the World Trade Organization in 2001 as a big driver of margin expansion, enabling cost-cutting as a huge lever to keep margins expanding. (It was also the year when worker productivity froze in its tracks.)
In the globalization regime, “you didn’t have to think too hard to make money and expand your margins,” she said. It was “very easy and fungible and frictionless” for companies to buy things from different places and contain costs. She also cited the low-interest-rate environment that persisted for much of the past few decades, enabling lots of “financial engineering.”
For example, Subramanian said it was common to see companies that knew they would miss their earnings estimates borrowing money and buying back stock to hit their targets, adding the caveat that “there are good reasons to do share buybacks and bad reasons to do share buybacks.” This all “really created a lot of bizarre behavior.”
Warren Buffett’s long-time fondness for stock buybacks has even come under fire from other investors, with Jeremy Grantham writing in 2023 that it facilitates stock manipulation and should be illegal. BofA Research found in July 2025, however, that stock buybacks had decelerated a bit, albeit they remained high by historical standards.
The situation now is harder in many ways, but companies aren’t able to financially engineer their way to earnings growth, she added. Now that’s a sea change.
One final note on the Shakespearean romances, from Drew Lichtenberg: that appellation came about in the late 1700s, nearly two centuries after Shakespeare’s lifetime, with the birth of the romantic movement. The word “romantic” had previously existed, but it didn’t have its current meaning until Samuel Taylor Coleridge elevated it to mean something that connects back directly to nature and the divine genius of humanity’s self-expression. This was largely a response to the Enlightenment’s elevation of reason and logic and its ultimate achievement: the Industrial Revolution that unleashed modern capitalism on the world. A sea change, indeed.
Wall Street responded sharply to Federal Reserve Chair Jerome Powell’s address at the Jackson Hole symposium on Friday. The speech, delivered amid mounting economic uncertainty and political pressure, provided new signals on the Fed’s interest rate outlook that immediately moved stocks, bonds, and global currencies.
Major U.S. stock indexes surged following Powell’s remarks, with the S&P 500 climbing over 1.5% and set to break a five-day losing streak—the longest since January. The Dow Jones industrial average and the Nasdaq Composite each rose by nearly 2%, buoyed by renewed hopes of a rate cut at the Fed’s next meeting in September.
Ten out of 11 S&P 500 sectors traded higher, led by health care and real estate, as traders interpreted Powell’s comments about weakening labor markets and persistent inflationary risks as supportive of easier monetary policy.
Bonds and currencies shift, Europe and Asia rise
In the bond market, Treasury yields fell and the U.S. dollar weakened as traders increased their bets on a September rate cut. The CME FedWatch tool showed a roughly 70% probability of a 0.25 percentage point reduction next month, slightly below earlier expectations. European and Asian equity markets likewise gained, with China’s CSI 300 index reaching multiyear highs, reflecting global optimism for lower U.S. rates.
Powell’s speech carefully balanced concerns: He acknowledged unusual weakness in both hiring and inflationary trends, and signaled confidence that the impacts from tariffs might be temporary. Investors and analysts viewed this as an indirect endorsement of policy easing, though Powell stopped short of explicitly pledging a cut.
Traders are now positioning for the Fed’s September meeting, with markets pricing in up to two rate cuts by year’s end. Analysts caution that sustained market gains are contingent on Powell’s continued focus on downside risks and the ongoing tug-of-war between inflation data and jobs figures.
Overall, Powell’s Jackson Hole speech quickly reversed the recent stock market slide and triggered renewed optimism that the Fed may act to support growth—while underscoring the uncertainty clouding monetary policy for the remainder of 2025.
For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.
Login
