The LLM-as-commodity thesis fails a… — resolves 2027-07-25

The LLM-as-commodity thesis fails a stress test against all three of its own watchpoints, which have moved the opposite way: annual vendor-switching rates near 11%, rising lab gross margins alongside flagship price increases, and enterprise share concentrating in Anthropic (32% up to 40%, with the top three now at 88% combined). A foundry analogy makes the same point: real foundry economics concentrate pricing power at the leading-edge tier while commoditizing the trailing edge, so commoditization is a line you climb rather than a switch that flips for an entire layer. That converges with a narrower reading in which frontier labs behave like integrated device manufacturers holding a ceiling over the application layer.

BG2 Pod / Altimeter(guest host)訪 Ali Ghodsi(Databricks CEO)+ Arvind Jain(Glean CEO)
「the the LLMs is a commodity as Arvind said. You ca…」

Stratechery(Ben Thompson)
「TSMC was unique precisely because they didn't desi…」

Stratechery(Ben Thompson)
「Claude Code alone has achieved an annualized recur…」

SemiAnalysis
「Anthropic's ARR has exploded from $9B to over $44B…」

SemiAnalysis
「AWS margins inflected this past quarter driven pri…」

AI Snake Oil / Normal Technology (Arvind Narayanan & Akash Kapur)
「If we assume a 5% net margin — much higher than ai…」

arXiv (Satoshi Matsuoka)
「a depreciation conveyor delivers newly amortized f…」

Reviewed over the next 12 months, through 2027-07-25: (1) Menlo Ventures' next enterprise report, expected around 2026-09, shows share flattening out, which would revive the commodity thesis; (2) Anthropic's audited gross margin, if disclosed, comes in well below the mid-60s range, which reopens the margin leg of this call; (3) flagship prices actually fall, or Claude Sonnet's end-August promotion does not revert to standard pricing; (4) a lab fails to bring a second adjacent application in-house within 12 months while a third-party app holds its ground on that turf, which would downgrade the IDM-ceiling reading; (5) once the current compute shortage clears, growth in the data/application layer outpaces the lab layer.

Jul 25, 2027

2026-09-30(Menlo 年中報告+Sonnet 8-31 回價)

Not due yet — no public outcome to report.

Not resolved yet — here is the condition we pre-committed to that would overturn this call (check it yourself, you don't have to wait):

Any of the following flips the call and revives the commodity thesis: Menlo's next report shows enterprise share becoming less concentrated rather than more; Anthropic's first disclosed gross margin comes in well below the mid-60s baseline; or flagship pricing is actually cut and the Sonnet promotion does not revert to standard pricing.