Nvidia shares climb as Meta, Amazon, Alphabet and Microsoft ramp AI capex, even as investors question the pace and payoff
Four US tech giants are planning roughly US$650bn in AI-focused capex by 2026 — a scale of spending that is reshaping tech valuations, driving a debt boom and putting suppliers like Nvidia at the centre of the trade, while investors question how quickly the cash will come back.
According to Bloomberg, Alphabet, Amazon, Meta and Microsoft together forecast about US$650bn in capex in 2026, largely for AI data centres and high-end chips.
Meta has guided full-year capex up to as much as US$135bn, Microsoft is projected to spend almost US$105bn this fiscal year after a 66 percent jump in Q2 capex, Alphabet has flagged up to US$185bn for 2026, and Amazon has laid out US$200bn in planned capex for that year.
The market reaction has not been uniformly positive.
Bloomberg reports that the four companies have lost more than US$950bn in combined market value since releasing their latest earnings and spending plans, even though many of their core businesses — from ads and search to ecommerce and productivity software — remain solid.
On the other side of the trade sit the suppliers.
According to CNBC’s Halftime Report,” Nvidia CEO Jensen Huang called the tech industry’s AI capex “justified, appropriate and sustainable,” arguing that “all of these companies’ cash flows are going to start rising.”
CNBC reported that Nvidia shares closed nearly 8 percent higher after his remarks.
Huang described this as “the largest infrastructure buildout in human history,” driven by “sky high” demand for compute, and said all Nvidia GPUs sold to date — including older A100s — are currently being rented.
Huang also highlighted how big customers are using Nvidia chips.
Meta is shifting from CPU-based recommendations to generative AI and agents; Amazon Web Services is using Nvidia-powered AI to refine product recommendations; and Microsoft is using Nvidia chips to enhance enterprise software.
He further pointed to OpenAI and Anthropic, saying “Anthropic is making great money. OpenAI is making great money,” and argued that “if they could have twice as much compute, the revenues would go up four times as much.”
Bloomberg reports that the buildout is already straining real-world capacity, with the four platforms competing for limited electricians, cement trucks and Nvidia chips from Taiwan Semiconductor Manufacturing Co., creating “bottlenecks.”
At the same time, funding needs are pushing more activity into credit markets.
Bloomberg says AI-related firms and projects tapped debt markets for at least US$200bn last year, with projections in the hundreds of billions for 2026, spanning blue-chip bonds, junk debt, private credit and asset-backed structures.
That scale of spending also raises macro and cycle risk.
Investor Tomasz Tunguz told Bloomberg that these “cash-generating machines” now “need that cash, and they need more of it, so they’re borrowing.”
He compared the AI boom to past investment frenzies that “don’t always end well,” even if they are “huge catalysts for the economy” on the way up.
Steve Lucas, CEO of Boomi, told Bloomberg that what is “spooking people” is the analyst narrative about how fast AI will disrupt businesses.
“I would not debate the potential of AI,” he said. “I would absolutely debate the time frame, and I would passionately debate the economics.”
