06-reference

stratechery earning spending weekly

Thu May 07 2026 20:00:00 GMT-0400 (Eastern Daylight Time) ·reference ·source: Stratechery ·by Andrew Sharp
stratecheryweekly-digestbig-tech-earningsagentic-business-modeljoanna-sterncapexai-infrastructure

“2026.19: Earning & Spending” — @Andrew Sharp (Stratechery weekly)

Why this is in the vault

Weekly thematic roundup with strong week-over-week continuity on the agentic-business-model thesis. The week’s Big Tech Q1 coverage (Google+Meta, Amazon, Microsoft+Apple) reads as one connected argument: hyperscaler capex is now >3x the inflation-adjusted Manhattan Project per quarter, and Microsoft has publicly conceded that per-seat SaaS pricing breaks under agents. The week’s individual pieces are already filed; this digest captures Stratechery’s own connective framing — Andrew Sharp explicitly stitches the four earnings updates into one narrative (“strategy that actually looks a lot more rational than the numbers sound”) which is itself the load-bearing data point for RDCO. The remaining digest items (Joanna Stern interview, Boston Celtics post-mortem) are sports/media; not load-bearing for RDCO.

Issue contents

Stratechery’s curator (Andrew Sharp, not Ben Thompson) explicitly highlighted three clusters:

  1. What We Learned from Big Tech’s First Quarter (Daily Updates + Tuesday Article)

    • Google Earnings, Meta Earnings — “Wall Street loved Google’s earnings, and hated Meta’s, even though the latter’s core business was more impressive. The difference is that Google is monetizing its investments now (and it might be all Anthropic).” 2026-05-04-stratechery-google-meta-earnings
    • Amazon’s Durability — Amazon looked behind in AI training era, well-placed in inference era. 2026-05-05-stratechery-amazon-durability
    • Microsoft Earnings, Apple Earnings — Microsoft unveils its new agentic business model (“seats plus consumption”); Apple confronts memory/chip shortages even as Mac benefits from AI. 2026-05-06-stratechery-microsoft-apple-earnings
    • Sharp’s framing: “first quarter CapEx was more than three times that of the Manhattan Project” — four of five megacaps continuing the pour.

  2. A Conversation with Joanna Stern — interview about her book Living With AI, career-change LLM use, AI in mammograms, and limits of LLMs (the “preying mantis pregnancy misdiagnosis” hook). 2026-05-07-stratechery-joanna-stern-interview

  3. What’s Next for the Celtics? (GOAT podcast + Sharp Text “The Wolves Are Why We Do This”) — NBA playoffs commentary. Note: the digest snippet in the inbox truncated this as “Boston [Dynamics?]” but the actual content is Boston Celtics, not robotics. Skipped — sports/media adjacent, no RDCO mapping. The Sharp Text companion piece on the Timberwolves is also pure-sports.

Other items mentioned (no separate filings):

Mapping against Ray Data Co

STRONG continuity-mapping, not net-new: this digest is mainly useful as the connective tissue showing Stratechery’s own week-over-week argument is consistent and reinforces RDCO’s 2026-05-06-stratechery-microsoft-apple-earnings take.

What the digest reframes that the individual filings don’t:

  1. The “first quarter capex >3x Manhattan Project” framing is Andrew Sharp’s distillation, not Thompson’s. Useful concrete number for Sanity Check / Variant C v3 economic-scale arguments — when explaining to non-tech-finance readers what hyperscaler AI capex means in absolute terms, this reference (Manhattan Project parity, 3x over) is a more legible analogy than “$190B Azure capex” alone. Worth holding for the 2026-05-06-stratechery-microsoft-apple-earnings#mapping-against-ray-data-co downstream Variant C v3 work.

  2. Stratechery’s own narrative compression: “the numbers sound less rational than the strategy actually is.” This is the Stratechery editor pushing back on the bubble-thesis frame from the OUTSIDE of Thompson’s own daily writing. RDCO’s own positioning on LLM commoditization vs hyperscaler durability (per 2026-05-04-karlmehta-llm-commoditization-intelligence-rails + 2026-05-05-stratechery-amazon-durability) lands closer to Sharp’s frame than the bear case. Worth calibrating Sanity Check’s economic-scale piece accordingly: the strategic logic is rational even when the numbers sound mad, and that distinction is the actual interesting thread, not “is this a bubble.”

  3. No new agentic-business-model angle beyond Wednesday’s Microsoft/Apple piece. The digest does not surface a new datapoint that would change Variant C v3’s framing. The Microsoft “seats plus consumption” framing remains the load-bearing artifact of the week and is already filed in detail.

Verdict: status-only. The digest’s connective reading is useful but doesn’t surface a new decision. The Manhattan Project comparison is a writing-resource for later, not an action item.