The State of AI
Nvidia Beats Back Bubble Fears With Record $68 Billion in Sales in Fourth Quarter ‘Computing has changed,’ CEO Jensen Huang says, citing agentic AI as driver of 94% profit surge
(…) Data center hardware—the chips and networking equipment that Nvidia sells to AI and cloud-computing companies—accounted for 91.4% of the quarter’s sales, or $62.3 billion, and the segment’s revenue grew slightly faster than the company’s overall sales.
“The simple way to think about it is, computing has changed,” Nvidia Chief Executive Jensen Huang said on Wednesday’s earnings call with investors. “In this new world of AI, compute equals revenues…I am certain at this point that we’ve reached the inflection point” where agentic AI is upending how business is done worldwide and selling AI tools is starting to generate real profits. (…)
A factor that will likely shape Nvidia’s fortunes will be the transition from AI-model training to inference, the process by which AI tools respond to queries. Training and inference require different types of computing, and as a result, different hardware.
Nvidia has for years dominated the training market with its graphics processors, known as GPUs—powerful chips capable of performing billions of simple tasks simultaneously. As more tech companies deploy AI tools in the real world, demand is expected to shift from training to inference, which relies more heavily on central processing units, or CPUs, a simpler type of data-center chip that more companies are capable of designing.
Last week, Nvidia announced a partnership with Meta that included its first major deployment of CPUs that aren’t connected in servers to GPUs, a sign that customers such as Meta need more inference-computing infrastructure to run their AI tools and other applications.
“It’s important to understand that inference equals revenues for our customers now,” Huang said. “As AI agents come into wider use, being able to quickly generate the computing tokens needed to operate them, customers have realized that their capital spending on Nvidia’s products leads to faster growth.” (…)
Nvidia said Wednesday that it expected $78 billion in revenue in the current quarter, significantly higher than the $72.9 billion predicted by analysts, and gross margins of 75%, slightly higher than Wall Street’s prediction. (…)
- Even CEOs worry about AI (Axios)
Fortune 500 CEOs ranked AI and “new technology” as the top risk to their industry, in a survey out this morning. (…)
Data: Conference Board and Business Council. Chart: Jacque Schrag/Axios
Your business can now get knocked off its axis by all manner of doomsday content. A viral report, a post on X or even an announcement by a former karaoke company turned trucking firm can send a stock a tumblin’.
AI or new technology was identified as a top concern by 60% of the 142 CEOs surveyed in early February by the Conference Board, a nonpartisan think tank, and the Business Council, an association of CEOs.
- It ranked third the previous quarter. At the end of 2025, geopolitical risk was the top concern (59%), followed by “cyber” (56%) and AI (53%).
- Since the Conference Board started asking CEOs about AI in 2024, this marks the first time the technology topped the list.
AI isn’t making the CEOs feel that bad, to be sure. CEO confidence overall jumped into positive territory from the previous quarter, rising 11 points to a score of 59 (anything above 50 is positive).
- In the last three months of 2025, Nvidia was name-checked in 234 earnings calls, up from 160 over the same period the previous year.
Data: AlphaSense. Note: Excludes Nvidia’s own investment calls. Chart: Axios Visuals
The $130 Billion Race for Companies to Get Their Tariff Money Back At least 1,800 companies have filed lawsuits seeking refunds from the government, following the Supreme Court’s tariff ruling last week
(…) Through Dec. 10, at least 301,000 importers were subject to the tariffs that were ultimately struck down, Customs and Border Protection officials said in a court filing. That total likely includes many businesses, but also some individuals who paid tariffs directly on goods purchased overseas, lawyers have said.
The task of handling the cases is falling to the Court of International Trade, a specialized New York City-based federal trade court that has plenty of experience with matters like this—though none of them have involved as many potential litigants or a price tag of this magnitude. (…)
In filings in one of the cases that ultimately went to the Supreme Court, the administration’s lawyers assured lower courts that companies could be “made whole through a refund, including interest” if the tariffs were ultimately ruled unlawful.
On Friday, an angry Trump criticized the justices for not including a clear directive in their opinion and told a reporter asking him if the administration planned to issue refunds: “It’s not discussed. We’ll end up being in court for the next five years.”
In a Sunday appearance on Fox News, Treasury Secretary Scott Bessent said the administration would look to the lower court for guidance. “It’s out of our hands, since it’s in the court, and we will follow the court’s orders,” he said. (…)
In lawyers’ optimistic assessments, it could take as little as a year, or two. Pessimistic guesses run considerably longer. (…)
Not everyone is willing to take the step of going to court. Kimberly Daniels, a Washington, D.C.-based customs broker at Mercantile Logistics & International Trade, said 20 of her clients are looking to get refunds of $2,200 to $7 million each in the tariffs at issue. Of those, only the largest, publicly traded firm was able to file a case; the others don’t have the financial resources to hire lawyers. (…)
- It’s payback time for Trump’s tariff fiasco The administration handing out refunds to Chinese companies will be a terrible look
(…) The refunds, on top of remaining and forthcoming tariffs, will thus surely become a political issue between now and the midterm elections in November. If Trump had any sense he would be ostentatiously pushing them through quickly, perhaps even labelling them a tariff dividend and hoping no one notices it’s not exactly the one he promised.
But although his administration has been reducing tariffs either through negotiation or unilaterally, Trump himself apparently simply cannot grasp quite how unpopular they have become.
Another potential source of friction is the mismatch between who in effect bore the cost of the tariffs and those who will get the refund. The money is handed back to the “importer of record” which paid them, but if that is a consumer-facing company — or indeed a wholesaler — which passed the cost on to its customers, the latter might feel they are morally if not legally owed money back. (…)
There may be even more righteous anger to come. Ryan Petersen, chief executive of the global logistics technology company Flexport, which is also offering a tariff refund service, says the US is very unusual in allowing foreign companies straightforwardly to act as importers of record. Flexport says their analysis of customs data suggests that the share of trade with China accounted for by Chinese importers of record jumped from 9 per cent before “liberation day” in April 2025 to 20 per cent by the end of the year.
Petersen says this reflects Chinese companies giving themselves the ability to misvalue imports to reduce tariff costs. It also means that the US government, while letting consumers take the hit, will be shelling out billions of dollars to a rising number of Chinese companies who are aggressively targeting the US market.
This will be extraordinarily bad optics. Trump always said that Chinese companies would pay the tariffs. In economic terms, this has largely turned out to be wrong, as the cost of duties has been passed on to domestic producers and consumers. But in an administrative sense it seems to have been increasingly correct.
If you had to precision-design a policy to showcase the Trump administration’s shortcomings, the IEEPA tariff saga would be it. It’s an illegal duty based on wrong-headed economics, it was ineptly designed and incompetently administered, sulkily reversed under belated legal duress and is giving a windfall to exactly the people it was designed to punish. It would take a heart of stone not to laugh, but it’s unlikely American consumers and voters will appreciate the joke.
Americans Are Leaving the U.S. in Record Numbers
In its 250th year, is America, land of immigration, becoming a country of emigration?
Last year the U.S. experienced something that hasn’t definitively occurred since the Great Depression: More people moved out than moved in. The Trump administration has hailed the exodus—negative net migration—as the fulfillment of its promise to ramp up deportations and restrict new visas. Beneath the stormy optics of that immigration crackdown, however, lies a less-noticed reversal: America’s own citizens are leaving in record numbers, replanting themselves and their families in lands they find more affordable and safe.
Since the Eisenhower administration, the U.S. hasn’t collected comprehensive statistics on the number of citizens leaving. Yet data on residence permits, foreign home purchases, student enrollments and other metrics from more than 50 countries show that Americans are voting with their feet to an unprecedented degree. A millions-strong diaspora is studying, telecommuting and retiring overseas. (…)
More than 100,000 young students are enrolled abroad for a more affordable university degree. In nursing homes mushrooming across the Mexican border, elderly Americans are turning up for low-cost care.
On a conference call last month hosted by Expatsi, a relocation company, almost 400 Americans signed up to learn how to move to Albania. The former Stalinist state offers a special visa allowing U.S. citizens to live and work there, with no tax on foreign income for a year, no questions asked.
“Previously, the Americans leaving were super-adventurous and well-credentialed,” said Expatsi founder Jen Barnett, a 54-year-old Alabama native who moved to Yucatán, Mexico, in 2024.
“Now they’re ordinary people, like me,” she said as she ticked through growth numbers. In 2024 the company organized three group scouting trips for clients; this year it will be 57, she said: “Our goal is to move one million Americans.” (…)
The U.S. experienced net negative migration—an estimated loss of some 150,000 people—in 2025, and the outflow will likely increase in 2026, according to calculations by the Brookings Institution, a public-policy think tank. The number could be larger or smaller because official U.S. data doesn’t yet fully capture the number of people leaving, Brookings analysts noted. The total in-migration was between around 2.6 and 2.7 million in 2025, down from a peak of almost 6 million in 2023.
The U.S. saw 675,000 deportations and 2.2 million “self-deportations” last year, according to data from the Department of Homeland Security.
A Wall Street Journal analysis of 15 countries providing full or partial 2025 data showed that at least 180,000 Americans joined them—a number likely to be far higher when other countries report full statistics. (…)
Relocation agencies say their new clients go far beyond young adventurers on European sojourns or their retiring parents. They include Midwestern small-business owners—architects, financial advisers and engineers—saving on healthcare costs by living seven time zones east of their clients. Middle-aged divorcées are looking for a fresh start and Americans on disability or social security are trying to stretch their benefits. (…)
Across dozens of interviews, U.S. expats described their motivations as a tangle of economic incentives, lifestyle preferences and disenchantment with the trajectory of America, citing violent crime, cost of living and turbulent politics. Trump’s re-election was a factor for many—although others voted for him. But the structural and societal shift runs much deeper. When Gallup asked Americans during the 2008 recession how many wanted to leave the U.S., the answer was one in 10. Last year: One in five. (…)
The number of U.S.-based academics seeking jobs overseas rose by more than a fifth last year, according to Times Higher Education, a U.K.-based provider of global education data. Most of them landed in Europe, where the EU has set aside 500 million euros to lure top scientists to the continent. Professors teaching abroad blamed the American right for slashing research funding, and the left for policing university speech.
International students coming to America fell by 17% last fall and is expected to decline more quickly in years to come—while the cohort of Americans obtaining a degree in Europe has doubled from 2011, rising 14% last year alone in the U.K., according to UCAS, the British university admissions service. (…)
Of the 12 American students the Journal spoke to for this story, studying across Spain, Scotland and England, only one planned to return to the U.S. (…)
Elsewhere in the WSJ:
“Breaking: Statue of Liberty reportedly spotted swimming back across the Atlantic. Said she ‘preferred the original terms and conditions,’” French Response replied in January to a pro-Trump account on X that had said France could be conquered “as an after thought” following a U.S. takeover of Greenland and Canada.
- The US Is Flirting With Its First-Ever Population Decline America’s population wasn’t expected to start falling until 2081. Trump’s immigration crackdown means it could happen as soon as this year.
If there’s one single consistent advantage the United States has carried since its founding, it is its ability to draw talent and expand its population. Now, as the country prepares to celebrate its 250th birthday and ponders its appetite for President Donald Trump’s crackdown on immigration, the US risks recording a historic and economic milestone decades ahead of schedule: Based on at least one respected estimate, 2026 may see the first real population decline in American history.
Even if that milestone doesn’t happen this year, there’s broad agreement among experts on both sides of the immigration debate that Trump’s second term is hastening a critical point — when net migration into the US stops offsetting the declining births and rising deaths that come with an aging native-born population. The more Trump cracks down on immigration, the sooner the US population plateaus or even shrinks.
A country’s population is an essential element of its economic mass. The shrinking population of China, which in 2025 recorded its lowest birth rate since Communist rule began in 1949, is one good reason it may never overtake the US as the world’s largest economy. Japan’s population peaked at 128 million in 2010, and its decline has dragged on growth for years. Europe’s worsening demographics have long fed its narrative of economic malaise.
The US has for years mostly stood apart from that conversation. In 2023, when the US Census last issued long-run forecasts for the population, the main prediction was that it would decline for the first time in 2081. But the way things are going, this year the US is at best poised to record a lower population growth rate than Germany, where an aging population has contributed to its reputation as the “sick man of Europe.” (…)
In the year prior to July 1, 2025, the US Census revealed this week that the population grew by only 0.5%, or 1.8 million people, its lowest growth since the pandemic. The main cause for the significant slowdown was a collapse in net migration to 1.3 million from a peak of 2.7 million in the year prior to July 2024.
In that most recent period, there were 519,000 more births than deaths, according to the new Census figures. That surplus is shrinking, however. By 2030 it’s likely to disappear altogether, making the US entirely dependent on immigration for population growth, according to the nonpartisan Congressional Budget Office. (…)
Recent work by researchers at the center-right American Enterprise Institute and center-left Brookings Institution suggests the US is already experiencing net negative migration. Diving into the data available on inflows and outflows of both legal and undocumented foreign-born workers, they calculated in a recent analysis that the US had a net decline in the immigrant population of 10,000 to 295,000 in all of 2025. That would still imply a tiny net increase in the overall population.
It’s this year, though, where the AEI/Brookings team’s estimate gets more interesting. They predict the US will have net immigration somewhere between a gain of 185,000 and a decline of 925,000 in 2026 — a prediction made before the US announced yet more new restrictions on legal migration at the start of the year.
The biggest contributor to the slowdown in net migration, the authors said, has been a reduction in new arrivals rather than the high-profile deportations now receiving media attention.
The AEI/Brooking researchers don’t go further and look at the overall effect on the population, but they acknowledge the math. If the low end of their 2026 prediction comes true and birth rates don’t spike in an unprecedented way, the US would have a decline of more than 400,000 in its overall population. Even at the midrange of that forecast, the country is at least flirting with a population decline. It seems increasingly possible that in 2026 “we could be at around zero or negative on population,” says Tara Watson, who directs the Brookings Center for Economic Security & Opportunity and is one of the study’s co-authors.
Since the US began taking censuses in 1790, such a decline has never been recorded, according to demographers. (…)
The Trump administration, meanwhile, has seized on other maximalist estimates, pointing to a decline of as many as 2 million people in the foreign-born labor force and claiming to have deported 622,000 people in 2025. If either is true, then the US population is likely already shrinking. Experts, though, have been challenging those claims, calling the first a misreading of the statistics and the second an overstatement. (…)
There were around 14 million undocumented immigrants living in the US before Trump’s deportation campaign began. That means, in principle at least, that a US committed to large-scale deportations and little or no immigration could see its population shrink multiple years in a row.
Still, some are holding out hope for a broader immigration compromise. Trump signaled this week that he was ready to talk after the uproar over ICE shootings in Minnesota spread into Republican ranks. Congress could, at some point, also wake up and tackle comprehensive immigration reform, however unlikely that currently seems. In a country with a foreign-born population of some 50 million people and growing worker shortages, proponents argue it only makes sense.
Republican Representative María Elvira Salazar of Florida, the author and co-sponsor of a bill that would allow millions of undocumented workers in the US to remain — albeit without the promise of citizenship — argues that Trump is uniquely positioned to shepherd a compromise. “He is a guy that comes from construction and hospitality,” Salazar said at a Brookings event in January. “He knows that we need those hands.” (…)
An AI Productivity Boom? Don’t Count Your (Productivity Data) Chickens
The three P’s of GDP growth
- Population
- Participation (proportion of the population actually working)
- Productivity
From Yale’s Budget lab:
If economic growth is strong and job growth is weak, some would tell us we are in the middle of a productivity boom with the promised benefits (and perils) of AI upon us!
But we shouldn’t be so fast to jump to conclusions as economic data is never that simple. We may in fact be seeing a productivity boom. But that is still to be determined. If we are seeing it, it is still unclear why.
There are three reasons why what we are seeing may not actually be a real jump in productivity—or an irreconcilable gap between economic growth and job growth.
First, productivity is noisy data (as you can see in the chart above) and this is an important point we must recognize at the offset. We shouldn’t overreact to one or even two quarters of data. Looking over several quarters, we can see that productivity growth has averaged about 2.2%. That is strong, but not unusually so (about where we were headed into the pandemic).
The noise in productivity is partly due to how it’s measured: productivity is what’s “left over” after all the other inputs have been accounted for (the residual). In other words, productivity growth includes both actual productivity and any measurement error. That’s why most economists prefer to look at it over a longer period of time.
Second, jobs growth in 2025 was quite low. We know that because we just got the annual revisions for the establishment survey (where jobs data comes from) which lowered the number of jobs we added in 2025—a totally typical part of the measurement process.
But for GDP growth in 2025, we’re still waiting for two things: 1) the advance, second, and third estimate for the fourth quarter in 2025 and 2) the annual benchmark revision for GDP which will occur in July. Note that any comparison of jobs data and GDP data for 2025 is comparing revised jobs data to unrevised and incomplete GDP data.
Third, jobs growth in 2025 was, as said, quite low. But GDP data has been weird in 2025 partly because of policy and behavioral swings around trade. If you look at job growth relative to private-domestic final purchases (PDFP, what economists sometimes refer to as “core GDP”, which strips out some of these sources of volatility, and is a better predictor of future GDP growth than GDP itself), it is still low, but not as low as it is relative to the GDP data.
It should be said that even if you trust the productivity data, and think we are seeing an increase in productivity, there are other explanations besides AI.
First, productivity can rise in response to compositional issues (see discussion from Ernie Tedeschi and Callum Williams). In the graph above, you can see productivity rising dramatically during Covid (and less dramatically in 2009). Did we all become more productive when we were stuck at home and others couldn’t go to work at all? You may be shocked to hear that we did not. Instead, since the people who lost their jobs were disproportionately low-wage workers (who show up as lower productivity in the data), productivity rose due to a compositional effect. (A similar thing happened in 2009).
One reason job growth in 2025 was so low was because of changes in immigration policy. If the people being removed from the labor force were lower productivity workers, that will show up as an increase in productivity even though the productivity of the workers who remain behind has not changed. In fact, the same thing skewing job growth down may be skewing productivity growth up.
Second, if you look at the productivity data, it appears that much of the boost is coming from capital utilization due to increased productive investment. That would be consistent with an increase in productivity due to AI. But its important to distinguish at this point it is people investing in AI not people becoming more productive by using AI.
Could this be the beginnings of an AI productivity boom? Maybe! But this is not the data we should hang our hat on.
If productivity growth swings back in the next reading (for the quarter of 2025), that shouldn’t make us assume that there are no productivity impacts of AI, and vice versa. Productivity is one of the most important economic concepts and also one of the hardest to measure—particularly in real time. We’ll have better luck tracking measures like real wage growth and changes in occupational composition to give us a signal.
But until we get a clear signal one way or the other—we shouldn’t put all our eggs in the productivity data release basket.
Also discussed in Fear the Fear