Hyperscaler capex near $780 billion is consuming a growing share of cash flow, a16z charts show
Andreessen Horowitz puts hyperscaler capex near $780 billion this year and tech at roughly 55% of US capital spending. Its charts also show free cash flow being squeezed, and the hyperscalers are relying more on debt to fund new chip capacity.
The Engineer · Build desk
What happened
- One a16z chart traces hyperscaler spending from about $241 billion in 2024 to $416 billion in 2025 and roughly $790 billion in 2026.
- Epoch AI estimated in June that cash capex at Microsoft, Amazon, Alphabet, Meta and Oracle was growing about 70% a year, against roughly 23% for operating cash flow.
- a16z says Nvidia A100 rental rates have held at or above their level at the start of the year, even as newer B200 systems draw strong demand.
- About 30% of S&P 500 companies report some quantifiable AI impact, but only about 2% report a tracked metric, according to a16z.
- a16z announced a $1.1 billion Machine Age Fund for AI's physical buildout on August 28, about a month before the September 30 report.
Why it matters
- What it costs cost Teams that rent capacity pay for prices that held up. A plan that assumed last-generation GPUs would get cheaper with age has been wrong on A100s this year.
- What it limits constraint New GPU capacity now depends on lenders as well as hyperscaler cash, so tighter credit can slow the supply that teams plan around.
- What doesn't add up contradiction The report's spending and adoption figures pull apart: hundreds of billions are committed while about 2% of US households paid for an AI service as of April.
Next story · Invest · Oct 7
Ray Dalio blames debt and rising rates for an AI bubble he says is close to burstingConfirmed2 publishers
On that chart's figures, spending grew about 73% from 2024 to 2025 and is on course to grow about 90% from 2025 to 2026 [21][22]. The report's other chart rounds this year to $780 billion and says annual spending could cross $1 trillion soon after [5]. The two charts are $10 billion apart for the same year [23].
Capex comes out of operating cash flow. Free cash flow is what remains after it [20]. For most of the boom, the hyperscalers' existing businesses generated enough cash to cover the build [16]. Epoch AI's growth rates put a date on the end of that. If they held, capex would overtake operating cash flow around the third quarter of 2026 [17]. Past that point free cash flow goes negative, and the shortfall has to be funded from reserves or from outside the business [24].
a16z's explanation for the A100 rental floor is that cheaper compute draws more use, so older chips stay busy while total demand grows [18]. Runtimewire points out that the value of installed hardware depends partly on how much demand expands before the next generation arrives [19]. For the rental figure to apply to a specific team, the rates a16z tracked would have to match the contract term and region that team buys. Runtimewire also notes the report does not establish that every chip, operator or data center keeps its value, and a16z says the story is still unfolding [10].
The coverage backs the physical-capital reading on where the money is going. According to a16z, capital is moving into semiconductors, power, networking, construction, debt markets and skilled labor [2]. For a product sold as the cloud, the bill looks a lot like a construction budget. Neither write-up reports power prices, grid connection times or wage figures from the report. The case that electricity and skilled labor will cap what teams can build is an inference from capital flows. The only price series in the coverage that touches a team's budget is GPU rental [9].
The firm's own disclosures say some of a16z's data comes from third parties and portfolio companies and has not been independently verified [12]. Runtimewire notes that the report's figure for tech's share of S&P 500 earnings growth this year through late August, about 76%, comes without the calculation or dataset behind it [3]. The author, David George, leads the firm's Growth practice [14]. His thesis matches the physical buildout the firm's new fund is set up to back [13].
In my view, flat prices for older GPUs are the right planning assumption for a team that rents, until a rental series turns down. A team that owns A100s sits on the other side of the same price.
What to watch
- Hyperscaler cash-flow statements around the third quarter of 2026, which show whether capex actually passed operating cash flow as Epoch AI projected.
- A100 rental rates over the next few quarters; a sustained drop would undercut a16z's claim that cheaper compute keeps older chips in use.
- Whether the share of S&P 500 companies reporting a tracked AI metric moves off about 2%.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption30
- Hype gap+20
- Incentives75
- Confidence60
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
The report describes hyperscalers' cash flow as helping finance semiconductor capacity, with debt increasingly joining the mix.
- [2]
According to a16z, the AI infrastructure boom has started pulling capital toward semiconductors, power, networking, construction, debt markets and skilled labor.
- [3]
a16z estimates technology contributed about 76% of the S&P 500's total earnings growth in 2026 as of late August; Runtimewire says the article does not lay out the calculation or underlying dataset alongside the claim.
- [4]
a16z estimates technology now represents roughly 55% of U.S. capital spending.
- [5]
An a16z chart puts hyperscaler capital expenditures at roughly $780B this year, with annual spending potentially crossing $1T soon after.
- [6]
One chart in the a16z report shows hyperscaler spending of roughly $241B in 2024, $416B in 2025, and toward roughly $790B in 2026.
- [7]
a16z's charts show capital spending consuming an increasingly large percentage of hyperscaler operating cash flow, with free cash flow being squeezed.
- [8]
Epoch AI, in June, looking at Microsoft, Amazon, Alphabet, Meta and Oracle, estimated aggregate cash capex was growing around 70% per year versus roughly 23% growth in operating cash flow.
- [9]
The report says rental rates for Nvidia A100 GPUs have held at or above their level at the start of the year, even as newer B200 systems draw strong demand.
- [10]
The report does not establish that every chip, operator or data center will retain its value; a16z itself says the story is still unfolding.
- [11]
a16z says nearly 30% of S&P 500 companies report some quantifiable AI impact, while only about 2% report a tracked metric.
- [12]
a16z's own disclosures say some data comes from third parties and portfolio companies and has not been independently verified.
- [13]
On August 28, a16z announced its $1.1B Machine Age Fund for AI's physical buildout.
- [14]
The report was published September 30 by David George, who leads a16z's Growth practice.
- [15]
a16z says that, as of April, roughly 2% of U.S. households paid for an AI service.
- [16]
For most of the AI boom, Microsoft, Amazon, Alphabet, Meta and their peers could largely finance the buildout using cash flows generated by their existing businesses.
- [17]
Epoch AI estimated that if those growth trends continued, capex would overtake operating cash flow around Q3 2026.
- [18]
a16z's explanation is that cheaper AI compute encourages more use, allowing older chips to remain useful while total demand grows.
- [19]
The value of today's infrastructure depends partly on how much demand expands before the next generation of hardware arrives.
- [20]
Operating cash flow is the cash the core business generates; capex is cash spent building long-lived infrastructure; free cash flow is basically what remains after subtracting those investments.
- [21]
Hyperscaler spending grew about 73% from 2024 to 2025 on the a16z chart.
- [22]
Hyperscaler spending is on course to grow about 90% from 2025 to 2026 on the a16z chart.
- [23]
The two a16z charts differ by about $10B for 2026 hyperscaler spending.
- [24]
If capex exceeds operating cash flow, free cash flow is negative and the shortfall must be funded from reserves or outside the business.
Sources
2 independent publishers whose own reporting we read for this story.
- runtimewire.coma16z says AI's next growth cycle runs through chips, power and robots
1 article · September 30, 2026
- theneuron.aia16z State of Markets 2026: AI Became a Capital Cycle
1 article · October 1, 2026
Topics and entities
Topics
- AI Adoption
- Hyperscaler cash flow and financing
- GPU depreciation
- AI Infrastructure Capex