Voter-backed limits and planned Democratic investigations could slow data-center development
Anthropic’s planned initial public offering is the biggest litmus test this year for the AI trade, and it is not looking encouraging so far, according to The Wall Street Journal. More news came from Anthropic on Tuesday: the giant AI lab is seeking a $1.33 trillion valuation, below the $1.5 trillion investors were anticipating just a week earlier, while demand for its shares has run below expectations. The report described the weaker-than-expected demand as part of a long-brewing capital crunch for AI.
Political pressure is building. The elections last Tuesday tilted the balance of power in Congress toward Democrats, many of whom are less laissez-faire about AI than their Republican counterparts, and AI-related stocks fell Thursday, erasing $2.5 trillion in market value. Democrats expected to chair powerful committees in the Senate and House plan to investigate the effect of data centers on electricity prices and to examine the proliferation of circular financing deals — developments that could slow U.S. data-center development, as could the numerous anti-data-center ballot measures voters approved last week.
Hostility to data centers in the U.S. has been building, and last week’s election results made clear it is a serious threat to Silicon Valley’s grand AI ambitions, the report said. Conversations about how to navigate an opposition movement that has significantly sharper teeth have grown in the tech world over the past week. Some companies are approaching governors-elect to ascertain what they plan to do, although the mood overall has been gloomy. At another level, the data-center debate will have to balance concerns about electricity prices against U.S. national-security interests. Local political opposition could imperil specific data centers, though a countrywide moratorium appears far-fetched, the report said.
Prominent tech investor Brad Gerstner said on a podcast that 35% of U.S. data-center projects planned for next year could be in jeopardy. The report also cited SoftBank Group’s loss in its most recent quarter, which came largely from charges on power-generation equipment for AI data-center projects in the U.S.
The financing strain also affects chip buyers. The Wall Street Journal said chip buyers have drained their cash, run up debt and are now trying to persuade wary investors to supply billions more to fuel spending. If that money does not arrive, data centers will not get built, regardless of what politicians say. That makes a successful Anthropic IPO vital, the report said — and makes Anthropic’s offering a better place to look for signals.
A separate disruption followed over Nvidia. In a revelation Sunday, CEO Jensen Huang admitted an affair with an executive at a Singapore-based data-center company who U.S. officials believe was passing intelligence to China. The scandal cast doubt on Huang’s position at Nvidia, which he has led since the 1990s. Nvidia’s stock fell about 4% Monday and declined further Tuesday, and a House panel asked Huang to appear later this month.
In an email to employees, two of whom are his own children, Huang wrote, “I owe my family and all of you an apology. There is no excuse for what I did.” The report said the email confessed to the affair. It also said the scandal would probably have little effect on the boom in the long run unless Huang had to step down. Politicians weighing any attempt to force him out would factor in the geopolitical implications: a forced departure could weaken Nvidia and possibly even play into China’s hands. Strengthened controls on chip exports are likely, according to the report, but that could actually benefit the U.S. AI ecosystem by making more chips available domestically.
Some current concerns are more likely to have a lasting impact than others, The Wall Street Journal said. It pointed to a cutting-edge model from China’s DeepSeek that emerged early last year, raised concerns that AI labs would not need as many chips to train frontier models and sent Nvidia’s stock plunging. The reaction proved premature, according to the report: DeepSeek’s advances were driven largely by copying the intelligence of the world’s cutting-edge models, and the efficiency gains did not dent demand for computing power. Big Tech’s pocketbooks stayed open, and Nvidia’s stock has risen 45% since.
The report also examined Nvidia’s valuation, which looks cheap — even cheaper since the company’s alleged spy scandal took the wind out of its shares this week. Normally, a company as fast-growing as Nvidia with a forward price-to-earnings ratio near 12 would be a bargain for investors. But successful companies often look cheap when they are about to take a big fall, the report noted: investors will not place a high value on future profits when those earnings appear to be in jeopardy, and in highly cyclical industries like Nvidia’s, undervaluation near the top of the cycle is par for the course. It is not clear whether the divergence between Nvidia’s strong growth trajectory and weak valuation is justified. The company could prove markets wrong and keep expanding profits longer than its valuation implies, but a big measure of doubt about the future is baked into its price.
The report said cybercriminals are also going after computing power as a new kind of digital asset. Hackers are targeting so-called API keys — a kind of password that allows access to AI models from leading American labs such as Anthropic and OpenAI. The keys allow hackers to use those models, running up charges that are billed to the companies they stole the keys from.
Taken together, The Wall Street Journal described an AI boom facing financing pressure, political resistance and security risks, while cautioning that not every recent shock will prove lasting. Its answer to whether the turmoil is noise or evidence of a more permanent shift is likely a bit of both. Anthropic’s IPO remains a concrete market measure because the report said a successful offering is vital to continued data-center construction.