As concerns mount that surging AI spending by technology giants could be inflating a market bubble, Nvidia CEO Jensen Huang has stepped forward to calm nerves, publicly defending investments that now routinely reach into the hundreds of billions of dollars.
Speaking in an interview on Friday, February 6 (US Eastern Time), Huang said the sharp rise in capital expenditure on AI infrastructure across the tech sector is reasonable, appropriate, and sustainable. He described the current wave of spending as “the largest infrastructure build-out in human history,” driven by what he called “extraordinarily high” demand for computing power.
Huang’s comments coincided with a broad rebound in AI-related stocks on Friday. Nvidia shares surged as much as 8.8% intraday to hit $187, before closing up nearly 7.9%, snapping a five-day losing streak. The rally helped the stock recover from its lowest closing level since December 17 last year, even though Nvidia still ended the week down roughly 3%.
His remarks came at a sensitive moment for markets. Investors have grown increasingly skeptical about whether massive, sustained AI investments can generate attractive returns, especially after new tools released by Anthropic reignited fears that AI could upend traditional software business models. That anxiety recently triggered heavy selling in technology stocks, with some Goldman Sachs analysts even comparing today’s software industry to newspapers disrupted by the internet in the early 2000s.
According to reports on Friday, recent earnings releases, forward guidance, and previously announced plans indicate that Nvidia’s key customers—Meta, Amazon, Google, and Microsoft—are collectively planning around $650 billion in capital expenditure for 2026, roughly 60% higher than in 2025. This spending level exceeds the GDP of many mid-sized economies, with a significant portion expected to be directed toward Nvidia’s chips.
As previously noted, fears over AI investment efficiency sparked intense sell-offs over the past week. FactSet data show that the combined market capitalization of Microsoft, Nvidia, Amazon, Alphabet, Meta, Oracle, and other tech giants has shrunk by approximately $1.35 trillion.
The volatility has been stark. After reporting a 66% surge in second-quarter capital expenditure that exceeded expectations, Microsoft shares fell 10% last Thursday, wiping out $357 billion in market value in a single day—the second-largest one-day loss for an individual stock in US market history. Meanwhile, after Amazon forecast that its 2026 capital spending would rise 50% year over year, the stock plunged nearly 10% intraday on Friday before closing down about 5.6%.
AI Companies Are Profitable, Spending Will Keep Rising
Huang emphasized that cash flows at these technology companies are set to increase, and that AI infrastructure construction will continue for another seven to eight years. He stressed that AI has already become “extremely useful and extremely powerful,” with adoption rates reaching “very high” levels.
“As long as people continue to pay for AI and AI companies can generate profits,” Huang said, “they will keep doubling, doubling, doubling, doubling.”
He cited concrete examples of how Nvidia’s customers are monetizing AI. Meta is transforming recommendation systems that once ran on CPUs into platforms powered by generative AI and intelligent agents. Amazon Web Services’ use of Nvidia chips is reshaping how the retail giant delivers product recommendations. Microsoft, meanwhile, is using Nvidia-driven AI to enhance its enterprise software offerings.
Huang also singled out OpenAI and Anthropic as leading AI labs that are already “making a lot of money.” Nvidia invested $10 billion in Anthropic last year, and Huang said earlier this week that Nvidia plans to participate heavily in OpenAI’s next funding round. “If they had twice the compute,” he noted, “their revenue would grow fourfold.”
He added that every GPU Nvidia has sold in the past—including A100 chips from six years ago—is currently being rented out, underscoring the persistence of demand for AI computing power. Unlike the early days of the internet, Huang argued, today’s infrastructure is not sitting idle.
A Record-Breaking $650 Billion Investment Wave
Reports indicate that Alphabet, Amazon, Meta, and Microsoft together expect to spend about $650 billion on capital expenditure in 2026, marking an investment boom without parallel this century. Each company’s annual budget is projected to approach or exceed the total spent over the past three years combined, with any one of them setting a new record for single-company annual capital spending over the past decade.
Meta has said its full-year capital expenditure could reach as high as $135 billion, potentially up nearly 87% year over year. Microsoft’s capital spending rose 66% to $37.5 billion in its second fiscal quarter ended December 2025, and analysts expect its 2026 fiscal-year capital expenditure to approach $105 billion. Alphabet guided toward the high end of $185 billion in spending, exceeding analyst expectations and outpacing the budgets of many US industries. Amazon, for its part, announced plans to invest $200 billion in 2026—a forecast that also triggered a sharp sell-off in its shares.
By comparison, compiled media data show that 21 other major US companies—including leading automakers, heavy machinery manufacturers, railroads, defense contractors, wireless carriers, logistics firms, and names such as Exxon Mobil, Intel, Walmart, and GE spinoffs—are expected to spend a combined $180 billion in 2026.
DA Davidson analyst Gil Luria said these four tech giants see the race to provide AI computing power as a winner-takes-all or winner-takes-most competition. “None of them wants to lose this race,” he said.
Wall Street Worries About Efficiency and Overcapacity
Market reactions to the spending surge have been mixed. Meta and Alphabet shares have risen, while Amazon and Microsoft have been punished. Since releasing their latest earnings and outlooks, the four companies’ combined market value has fallen by more than $950 billion.
Paul Markham, investment director at GAM Investments, said markets are being swept by a “contagion effect of sentiment.” He warned that enormous capital requirements for large language model infrastructure, long payback periods, and the risk of overcapacity have become structural pressures weighing on investor confidence.
Theory Ventures investor Tomasz Tunguz observed that these tech giants were once “cash machines,” but now suddenly need vast amounts of capital—and more of it—forcing them to borrow. Media estimates suggest AI-related companies and projects raised at least $200 billion in debt markets in 2025 alone, with bond issuance tied to AI expected to reach several hundred billion dollars in 2026.
Tunguz has previously compared the AI boom to past investment manias, noting that while they do not always end well, “on the way up, they are massive catalysts for the economy.”
Steve Lucas, CEO of data and software integration firm Boomi, offered a more cautious view: “I wouldn’t question AI’s potential, but I would absolutely question the timeline of its development, and I would strongly question its economic efficiency.”
Execution risks also loom large. As data center construction accelerates, companies are competing for scarce resources—from electricians and cement trucks to Nvidia chips produced at TSMC facilities. “There are bottlenecks already, and there will be more,” Luria said.
Huang Also Sought to Calm Software Stocks Earlier This Week
This was not Huang’s first attempt this week to soothe market anxiety. On Tuesday evening, after software stocks were sold off, he spoke at a Cisco event and called the sell-off “the most illogical thing in the world.”
He argued that software products are tools, and AI will use those tools rather than replace them entirely. “There’s this idea that tools are declining and being replaced by AI,” he said. “Do you use a screwdriver, or do you invent a new screwdriver?”
Huang added that Nvidia itself relies heavily on such tools, freeing employees to focus on what the company does best: designing semiconductors and computer systems.
Software-related stocks fell for a second straight day on Tuesday as investors worried that tools released by AI model developers like Anthropic could eventually automate large portions of internal corporate work. A US software ETF dropped 4.1% that day to its lowest level since April, with AppLovin and Unity Software among the hardest hit.
JPMorgan analyst Toby Ogg wrote that investor appetite for technology remains weak. “We are now in an environment where the sector is not only guilty until proven innocent,” he said, “but is being sentenced before trial.”
Still, some analysts questioned the severity of the sell-off, arguing that software companies providing mission-critical tools are difficult to disrupt. Jefferies analyst Brent Thill pointed to Intuit, noting that its proprietary data and systems for navigating complex US tax laws give it a durable advantage relative to AI.
