The biggest technology companies on Earth are spending money at a pace that strains comprehension. After first-quarter 2026 earnings closed on April 29, the combined capital-expenditure plans for Amazon, Alphabet, Meta, and Microsoft landed at roughly $725 billion for 2026 — up 77 percent from 2025's record and already rising. An investor briefing circulating among money managers frames this as "the foundational infrastructure of the next economic era," not a bubble, and marshals specific figures to make the case. Several of those figures are wrong, and the errors all lean in the same direction.
That pattern matters to anyone using the memo to size a position, advise clients, or assess the risk profile of a bet on AI infrastructure. Getting the numbers right doesn't change the bull case's core logic — that spending is forward-looking and AI revenue is compounding fast — but it does change its strength.
The Capex Total the Memo Understates by $100 Billion
The briefing puts combined 2026 capital expenditures for the four hyperscalers at "more than $630 billion." That figure was defensible in late 2025. By the time Q1 2026 earnings closed, the consensus tally had reached approximately $725 billion — roughly $100 billion higher.
Meta raised its full-year guidance to $125–$145 billion, up $10 billion at both ends, citing higher memory-chip costs and additional data-center construction. Alphabet guided to $180–$190 billion, Microsoft to $190 billion, and Amazon held its earlier $200 billion commitment — CEO Andy Jassy telling investors the plan remained "largely the same." Jefferies analyst Brent Thill, surveying those results for the Financial Times, called the bear thesis "garbage" and said recent revenue growth justified the outlays. The memo understates the commitment at the center of its argument by roughly $100 billion — a gap that simultaneously makes the bull case look smaller than it is and makes the cash-flow pressure look less acute than it is.
Amazon's Free Cash Flow: The Number the Memo Actually Got Right
The briefing claims Amazon's free cash flow "fell from $38 billion down to $1.2 billion within a twelve-month period." This claim stands up. Amazon's Q1 2026 earnings filing, submitted to the Securities and Exchange Commission on April 29, states explicitly: "Free cash flow decreased to $1.2 billion for the trailing twelve months, driven primarily by a year-over-year increase of $59.3 billion in purchases of property and equipment, net of proceeds from sales and incentives. This increase primarily reflects investments in artificial intelligence." The comparable trailing-twelve-month period ending Q4 2024 shows free cash flow of $38.2 billion — a 95 percent decline.
The figure uses trailing-twelve-month windows, not calendar-year figures. On a calendar-year basis, Amazon's full-year 2025 free cash flow was $11.2 billion, down from $38.2 billion in 2024 — a 71 percent drop. Both figures confirm the same structural reality: Amazon's $200 billion 2026 capex plan runs against operating cash flow of approximately $139.5 billion, making negative free cash flow for the year mathematically unavoidable. Analyst estimates for the full-year 2026 deficit range from $17 billion to $28 billion. Amazon's long-term debt has already risen to $119.1 billion, up from $65.6 billion a year earlier.
Bank of America calculates that hyperscaler capital expenditures now consume roughly 94 percent of operating cash flow after dividends and buybacks, leaving the group reliant on external financing. The five largest hyperscalers raised $108 billion in bonds during 2025, a figure confirmed by multiple financial analyses tracking the debt surge.
Where the Revenue Gap Is Wider Than the Memo Allows
The briefing frames a "disconnect" between "$100 billion to $150 billion in annualized AI software revenue" and the capex commitments, then argues the gap is "rapidly normalizing." The top of that range reflects one of the most generous possible revenue definitions — and the memo presents it as if it were settled consensus.
Microsoft's Q3 fiscal-2026 earnings release provides the most specific disclosure available: CEO Satya Nadella confirmed the AI business "surpassed an annual revenue run rate of $37 billion, up 123% year-over-year." That figure covers clients running AI workloads on Azure, model-building companies, and Microsoft's own AI applications. Amazon Web Services reported an AI revenue run rate above $15 billion in Q1 2026. Google Cloud grew revenue 63 percent year over year to $20 billion in Q1 2026, with backlog nearly doubling quarter-on-quarter to over $460 billion, according to CEO Sundar Pichai. Meta's AI investment is largely internal, flowing to its Andromeda retrieval system, Lattice ranking model, and GEM generative-recommendation platform, which have lifted ad quality and conversion rates rather than generating external revenue.
The $25 billion figure that analysts have cited for direct AI-specific services revenue dates from a narrower, earlier definition that excludes most cloud growth attributed to AI workloads. The memo's "$100–$150 billion" range is reachable only under the most expansive attribution. Neither figure, however, closes the gap against $725 billion in annual capex spending. The memo softens precisely that gap by quoting the optimistic range as if it were consensus, when in fact the range is contested and definition-dependent.
The Depreciation Load Nobody Is Highlighting
The bull case has genuine support. Google Cloud's 63 percent revenue growth, its $462 billion backlog, and the 7 percent after-hours surge in Alphabet shares on April 30 are not fabricated. AWS grew 28 percent in Q1 2026 at a 37.7 percent operating margin — its fastest growth pace in fifteen quarters. The hyperscalers are profitable operating companies, unlike the dot-com cohort, and demand signals — extended GPU contract lead times, capacity constraints running into 2026, and signed multi-year compute commitments from OpenAI and Anthropic — are concrete.
The same earnings season also surfaced the risk the memo skirts. AI hardware assets depreciate at roughly 20 percent per year; at the current $725 billion combined spending pace, the implied annual depreciation load approaches $400 billion — more than the group's combined 2025 profits. Two-thirds of Microsoft's most recent quarterly capex went to short-lived GPU and CPU assets that begin compressing operating margins within months of deployment, not years. Meta fell approximately 6 percent after raising its capex guide, and short-seller Michael Burry has argued the group understates annual depreciation by $176 billion over 2026–2028 by applying five-to-six-year GPU life schedules against an economic life closer to two to three years.
There is also a structural circularity in the books that the memo does not address. Three of the four hyperscalers recorded large non-cash gains this quarter from their stakes in AI labs: Alphabet booked $36.8 billion in equity gains, mostly a markup on Anthropic; Amazon recorded $16.8 billion in pre-tax gains on the same investment; Microsoft recognized $5.9 billion in OpenAI-related gains over nine months. Amazon's Q1 2026 net income of $30.3 billion includes $16.8 billion in pre-tax gains on its Anthropic stake — a non-recurring item that flatters the headline figure.
The Honest Frame
The investor briefing is right that the build-out is forward-looking and that AI revenue is compounding at speed. AWS at 28 percent growth with a 37.7 percent margin, Google Cloud at 63 percent growth with a $462 billion backlog, and Microsoft's AI business at $37 billion in annual run-rate revenue and growing — these are not the numbers of a sector that has failed to monetize.
But a briefing whose specific errors all lean optimistic — understating capex by $100 billion, citing an inflated revenue figure as if it were settled, and drawing a "not a bubble" conclusion from evidence that cuts both ways more sharply than the memo acknowledges — warrants the extra scrutiny that its directional uniformity invites. The verified story is more instructive than the smoothed version: a $725 billion annual commitment, depreciation approaching combined profits, Amazon heading toward its first year of negative free cash flow, and a market that rewarded Alphabet for demonstrated monetization while punishing Meta for spending without an equivalent revenue story. That is not a bubble call. It is also not a sure bet.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
