NEW YORK, Feb 8, 2026, 07:39 EST
Amazon.com sketched out a $200 billion capital spending plan for 2026, with the bulk aimed at assets like data centers and servers. The stock fell 9% on Friday. That spike in outlays is stirring up echoes of the dot-com era’s infrastructure binge in the early 2000s—an effort that gave birth to the current internet but ended up handing backers only modest gains.
America’s top tech names are on track to sink over $630 billion into AI data centers and chips this year, even though those heavy bets haven’t delivered matching returns. Morgan Stanley analysts cautioned that investors “are not forgiving” when companies keep spending big without demonstrating clear returns on invested capital. Google Cloud’s revenue jumped 48% in the December quarter, well ahead of AWS at 24% and Microsoft’s Azure at 39%, sustaining the scramble to expand infrastructure. Reuters
Cash is suddenly in the spotlight. According to The Information, Amazon, Google, and Meta Platforms may see their free cash flow nearly vanish as their capital expenditures surge. That could leave them weighing whether to rein in share repurchases or take on more debt. Still, the report points out, these tech giants have enough balance sheet room to tack on hundreds of billions in debt if they want.
On earnings calls this week, Alphabet mapped out plans to double its capex for 2026, eyeing close to $185 billion. Meta put its full-year capital spending forecast as high as $135 billion. “We’ve never invested this much in anything before,” said Gil Luria, who heads up technology research at D.A. Davidson. The crunch, according to Jefferies analyst Brent Thill, isn’t chips or GPUs anymore—the logjam is power and buildings. “Now, it’s a physical shell shortage,” he said. Fortune
Not all AI stocks are moving in lockstep. Amazon dropped again in New York and London on Friday following its latest spending plans. Alphabet and Meta, too, finished lower. “It’s not that the trade is over, but it got too pricey,” said Andrew Wells, chief investment officer at SanJac Alpha. He described the move as a “de-risking trade.” Nvidia, on the other hand, gained, with CEO Jensen Huang calling demand “sky-high” and describing the spending increase as “appropriate and sustainable.” Reuters
The divide is playing out sector by sector. Shares of ServiceNow dropped 12% this week, while Salesforce slid 9%, Reuters said, as investors start to doubt how well software companies can protect their margins when customers gravitate toward AI-focused tools. “It is a signal that investors are differentiating between who enables AI and who may be disrupted by it,” Charu Chanana, chief investment strategist at Saxo, wrote in a note. Reuters
AI’s impact is spilling into the physical world, too. Amazon, Google, Microsoft, Meta, and Oracle together are on pace to pour some $700 billion into AI-centric projects this year. That firehose of spending is starting to strain supply outside of Silicon Valley. Last week, Apple flagged to investors that it’s struggling to secure enough of two key chip types for iPhones and Macs. According to IDC analyst Francisco Jeronimo, phone and PC vendors could end up hiking prices by 5% or more this year, shipping less advanced models—or possibly both. Apple CEO Tim Cook declined to guess whether iPhone prices might rise. “That alone should give everyone pause,” investor Roger McNamee wrote in an email. The Washington Post
Contractors aren’t immune. The Associated Builders and Contractors says the sector faces a need for 456,000 fresh hires in 2027, climbing from 349,000 required this year. “Failing to do so will worsen labor shortages … upward pressure on labor costs,” warned ABC’s chief economist, Anirban Basu. New data center construction spending soared 32% in the first ten months of 2025 compared with the same stretch a year ago, according to the group. Fortune
The surge in spending isn’t without its pitfalls: costs are landing now, but returns are still murky. Delays can pile up from power hookups, permits, or finding the right workers. There’s another catch—AI demand might not be as steady as projections suggest. If the expected revenue doesn’t materialize soon, the same cash crunch prompting all these buy-or-borrow calls could trigger a rapid pullback in outlays.
