Amazon's $100B AI Capex Nobody Is Talking About

Amazon enters the second half of 2026 in a peculiar position: it is, by a comfortable margin, the largest artificial-intelligence infrastructure spender on the planet, and yet Wall Street is paying it essentially no narrative premium for the privilege. On a 8 June 2026 broadcast of TBPN, co-host John laid out the arithmetic using public.com data and his own reading of capex guides: Amazon is on track to spend more than $100B on capex this calendar year, ahead of Microsoft's roughly $90B, Alphabet's $75B and Meta's $65B — a combined hyperscaler bill of about $350B. "Amazon has been weirdly out of the headlines on all things AI capex… obviously they have the actual core data center infrastructure and cloud leadership, but it doesn't feel like they're part of the conversation right now," co-host Jordy added.

That disconnect between capital deployed and narrative credit is the story.

The quarter that produced the headline numbers was, on paper, a blowout. Amazon posted Q4 net sales of $187B, up 10% year over year, with net income jumping 88% to $20B. But the reaction told a different story. Q1 guidance came in at $151–155B in net sales and $14–18B in operating income, with the high end of the range landing below Street consensus. AWS, the original cloud franchise, grew 19% to $28.7B — solid, but a touch light against a market that had just digested soft cloud prints from Alphabet and Microsoft. The stock fell roughly 3% on the week despite being up 40% over the prior year.

For a company spending more on AI infrastructure than any of its three largest competitors, that price action is the puzzle.

The narrative gap

Part of the explanation is that Amazon sells infrastructure, not stories. Microsoft has OpenAI. Google has Gemini. Meta has Llama. Each of those names is a recognizable AI brand that an institutional salesforce can put on a slide. Amazon has Trainium, Bedrock and a quiet lead in custom silicon, but no equivalent consumer-facing mascot. The result is that AWS, despite being the largest and oldest cloud, gets priced more like a low-margin utility than a software platform.

Jordy's framing was pointed: Amazon lacks the consumer AI narrative of its peers. Microsoft has the partnership story with OpenAI; Google has Gemini shipping in Workspace and Pixel; Meta is open-sourcing its frontier models and collecting goodwill from the developer ecosystem. Amazon has… data centers.

The bearish read would say the spend is defensive: hyperscaler capex is converging toward commodity economics, and Amazon is merely trying to keep pace. The bullish read — and the one the capex numbers lean toward — is that Amazon is buying optionality at scale while the rest of the industry is buying narratives.

The structural tailwinds the market is ignoring

Two political-regulatory shifts tilt Amazon's competitive position sharply in 2026, and neither is in the stock chart yet.

First, the de minimis suspension. For years, Shein and Temu used the low-value-import exemption to ship packages under $800 into the United States duty-free, undercutting Amazon's third-party sellers on price. The Trump administration's suspension of that provision removes the structural pricing advantage that had been hollowing out Amazon's marketplace. Combined with rising ad fees and a tightening trademark-enforcement regime, the third-party platform is suddenly less attractive to Chinese knockoff merchants and more attractive to brand-holding American sellers.

Second, the FTC retreat. Lina Khan's departure and the Trump-era antitrust easing mark a meaningful change in posture toward the company's rollup strategy. The agency that spent 2023–2024 preparing a monopolization case is, in 2026, more focused on conduct than structure.

John was blunt on the third-party side: "We don't have that and then we don't have, on the retail side, all the sellers are upset. No seller on Amazon is saying I'm so happy I built my entire business on Amazon." He cited analyst Sean Frank's argument that e-commerce deal volume is down roughly 50% from 2021, that Amazon Marketplace sales are flat-to-down, that the rollup acquirers (Thrasio, etc.) have gone bankrupt, and that fees have risen so much that doubling revenue often leaves EBITDA unchanged. Solo Stove, he noted, peaked at a $2B valuation and is now worth under $100M.

That critique is real. But it is the critique of the seller economy, not the platform economy. The platform's take rate has never been higher, and the de minimis change shifts the mix of who can afford to use it.

The generalist money desk is changing too

While the Amazon debate consumes the airtime, the venture industry is quietly going through its own structural reset. On the same 8 June broadcast, John walked through Eric Newcomer's October 2024 scoop on General Catalyst's Fund 12: $8B closed across four vehicles, with $1.5B in Creation (venture buyouts, 25% ownership targets), $1.5B in Ignition (seed through Series B, 10–25% targets), $750M in Health Assurance and $2.25B in Endurance for growth and late stage.

The economics have moved decisively away from 2-and-20. Carry now starts at 25% and climbs to 30% above 133% of contributed capital. Management fees range from 1.75% on Endurance to 2.5% on Creation and Health Assurance. On Fund 12 alone, the firm pulls roughly $120M a year in management fees — and that's before the fund-of-funds platform for emerging managers and the wealth-management product the firm has layered on top.

The carried-interest fight is the political variable. On Manifold Markets, the odds of repeal by end of 2026 sit at 40% yes, 60% no. The reported Trump administration proposal would reclassify fund-manager carry as ordinary income rather than capital gains. Dan Primack, the hosts noted, told a source the move was "like the Red Wedding."

The likely outcome, the hosts speculated, is something narrower: an offshore-only application designed to push capital back into US-domiciled investments rather than a full kill shot. Either way, the structural math for funds above $10M changes dramatically under ordinary-income treatment. General Catalyst's response — multi-vehicle diversification, tiered carry, fund-of-funds and wealth — looks less like a portfolio choice and more like a hedge against a regulatory regime where the standard partnership structure is no longer the obvious wrapper.

The capex story, restated

Strip the politics out and the Amazon thesis is straightforward. The company is deploying more capital into AI infrastructure than any of its competitors. It is doing so from a balance sheet that printed $20B of net income in a single quarter. It is operating in a regulatory environment that is, for the first time in four years, tilting in its direction on both antitrust and trade. And it is doing all of this while trading at a multiple that does not reflect any of it.

The bear case has a real answer: AWS growth at 19% is a deceleration signal, not a reacceleration. The third-party seller economy is structurally challenged. The consumer AI narrative gap is real and may persist. Q1 guidance was soft. None of those facts is in dispute.

But a $100B-plus AI infrastructure build, in a year when the de minimis loophole that threatened the marketplace is closed and the FTC is no longer the structural risk it was twelve months ago, deserves a different multiple than a regional cloud reseller. The market will eventually notice. The question is whether it notices before or after the next earnings call, when the capex guide ticks up again.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://www.youtube.com/watch?v=RR0OOIwIrHE