Meta, Amazon, Microsoft and Alphabet are expected to devote roughly $750 billion to capital expenditures this year alone, Bloomberg Businessweek reported on 12 October, much of it on data centres for artificial intelligence.
The companies’ own July outlooks add up to about that: Amazon about $220 billion, Alphabet $195 billion to $205 billion, Microsoft about $175 billion for calendar 2026 and Meta $130 billion to $145 billion. The sum, $720 billion to $745 billion, is our arithmetic.
Business Pill · CAPEX AND DEPRECIATION
A one-minute explainer of capex and depreciation: how money spent on long-lived assets becomes a cost over years. It teaches the general idea only and says nothing about these four companies.
The key insight: As we read it, the headline number is the sum of four separately stated plans: Bloomberg’s roughly $750 billion lines up with the companies’ own July outlooks, which add up to $720 billion to $745 billion by our arithmetic.
What Bloomberg Reported
Bloomberg wrote that the roughly $750 billion is “more than the entire gross domestic product of Ireland” or Belgium, more than half of what all S&P 500 companies spent in total last year, and “enough to cut a $5,500 check for every household in the US.”
John McWilliams of the real estate services firm Cushman & Wakefield told Bloomberg he would once have found it impossible to imagine four companies spending that much in a year. “We’ve never seen it,” he said.
Bloomberg said the cost of construction materials such as steel and cement has risen sharply, and that the price of short-term memory used in data centres is more than five times what it was a year ago, according to data compiled by Bloomberg Intelligence. It said suppliers of power equipment have backlogs stretching beyond 2030.

What Was Posted on X
On 17 October the Unusual Whales account posted Bloomberg’s estimate that Meta, Amazon, Microsoft and Google are expected to spend $750 billion on capital this year. The $5,500-per-household comparison is Bloomberg’s; the post’s ‘up more than 80% from $410 billion’ comparison is the poster’s own.
On 18 October the Fireside Alpha account posted a clip of Sequoia’s Alfred Lin arguing that AI founders shouldn’t worry about defensibility because large companies are distracted: โTheyโre trying to spend as much of their free cash flow as possible to build data centers right now. Theyโre not actually building the feature that youโre building.โ The view is Lin’s; the post quotes an earlier Fireside Alpha clip of Conviction’s Sarah Guo making a similar point about incumbents’ A-teams.
What the Companies Guided
Each figure below is the company’s own 2026 outlook from its second-quarter results in July, as stated or as reported from its earnings call.
Meta said in its 29 July release: “We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of $130-145 billion, narrowed from our prior outlook of $125-145 billion.”
On Alphabet’s 22 July call, chief financial officer Anat Ashkenazi said the company was updating its full-year 2026 capex guidance to $195 billion to $205 billion, up from $180 billion to $190 billion, and that it expects capex “to increase significantly in 2027.”
Amazon chief executive Andy Jassy said on the company’s 30 July call that Amazon now expects to spend about $220 billion in cash capex this year, up from roughly $200 billion, attributing the increase to the higher cost of memory chips, GeekWire reported.
For Microsoft, whose fiscal year ends in June, CFO Dive reported that a lease-accounting change adjusted the company’s calendar-2026 capex expectation to about $175 billion from a previously estimated $190 billion. “Outside of this useful life impact, our calendar year 2026 CapEx investment expectations remain unchanged,” chief financial officer Amy Hood said, according to CFO Dive.
What One Data Centre Costs
Bloomberg used a theoretical one-gigawatt AI data centre in the Abilene, Texas area to show where the money goes. Cushman & Wakefield supplied the costs of the shell and Bloomberg Intelligence estimated the hardware; on the high end of that analysis, Bloomberg put the total at roughly $30 billion.
Bloomberg said more than 60 gigawatts of such capacity is expected to be added in the US over the next four years, tripling current levels.
Bloomberg also noted that Michael Burry and Ray Dalio have warned of an AI bubble. It quoted Burry as saying that “All the capital being spent and lent by the richest companies on Earth” won’t be enough to keep the bubble from eventually bursting.

Why It Is Back on Our Radar
The figure circulated again on 17 October, when the Unusual Whales account posted Bloomberg’s $750 billion estimate and the $5,500-per-household comparison, which is Bloomberg’s. The post added that the total is up more than 80% from $410 billion last year; that comparison is the poster’s, and the Bloomberg text we read does not state it.
It came in the same week that, as we reported, the Philadelphia Semiconductor Index fell for an eighth straight session and CoreWeave’s 2032 bonds yielded close to 15%.
The Structural Read
As we read it, Amazon and Alphabet account for the larger share on these outlooks: about $220 billion and $195 billion to $205 billion, against about $175 billion for Microsoft’s calendar year and $130 billion to $145 billion for Meta.
As we read it, three of the four outlooks moved up through the year: Meta narrowed its range from $125-145 billion and Alphabet raised its from $180 billion to $190 billion, while GeekWire reported that Amazon’s rose from roughly $200 billion. Microsoft’s calendar-2026 figure went down, from about $190 billion to about $175 billion, which CFO Dive attributed to a lease-accounting change.
As we read it, Bloomberg’s $30 billion one-gigawatt example shows the scale per site: on that estimate, $750 billion is the order of what 25 such sites would cost, which is our arithmetic and not Bloomberg’s.
John McWilliams, Cushman & Wakefield, to Bloomberg Businessweek, 12 October 2026
“We’ve never seen it”
Three Implications
THE SUPPLIERS Bloomberg says suppliers of power equipment have backlogs stretching beyond 2030 and that data-centre memory costs more than five times what it did a year ago.
2027 Alphabet said on its July call that it expects its capex to increase significantly in 2027.
THE COMPARISONS The four outlooks measure capex differently (finance leases, cash capex, a lease-accounting change), so the combined total is approximate.
The Business Engineer Lens
This story maps onto the Business Engineer framework The AI Capex Map & The State of AI Hyperscalers.
The framework puts it this way: “This is not a continuation of the cloud capex cycle. This is a different cycle, with different physics.”
As we read it, this week’s figures sit inside that frame: the framework was published in May, when it cited a $725 billion combined 2026 figure, and the companies’ July outlooks sum to $720 billion to $745 billion. The framework does not discuss Bloomberg’s October feature.
What Is Not Established
The four outlooks are not measured the same way: Meta’s includes principal payments on finance leases, Amazon’s is described as cash capex, and Microsoft’s calendar-year figure reflects a lease-accounting change. The $720 billion to $745 billion sum is our arithmetic and does not adjust for those differences.
We did not read Microsoft’s or Amazon’s earnings call transcripts directly; their figures come from CFO Dive’s and GeekWire’s reports of the calls. We did not contact the companies.
The Bottom Line
Bloomberg reported that Meta, Amazon, Microsoft and Alphabet are expected to devote roughly $750 billion to capital expenditures this year. Their own July outlooks, as stated or reported, sum to $720 billion to $745 billion by our arithmetic.
Bloomberg put the cost of a single one-gigawatt AI data centre in the Abilene area at roughly $30 billion.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The roughly $750 billion figure is Bloomberg Businessweek’s, from its 12 October feature; we read its text but not its interactive cost breakdown. Meta’s figure is from its 29 July release and Alphabet’s from its 22 July earnings call; Amazon’s and Microsoft’s come from GeekWire’s and CFO Dive’s reports of their calls. The $720 billion to $745 billion sum and the chart’s midpoints are our arithmetic.
The 17 October post embedded here is by the Unusual Whales account; its comparison with last year is the poster’s. We did not contact the companies.
Sources: Bloomberg Businessweek (12 Oct 2026) · Meta Q2 2026 results (29 Jul 2026) · Alphabet Q2 2026 earnings call (22 Jul 2026) · GeekWire on Amazon’s Q2 call (30 Jul 2026) · CFO Dive on Microsoft’s Q4 FY26 call (30 Jul 2026)









