SemiAnalysis

SemiAnalysis published 39 pieces between 2026-05-01 and 2026-09-07, covering 15 tickers plus 4 macro pieces with no single ticker. Quant Data extracted 1,056 discrete claims from them, of which 13 were directional statements on a named ticker. Over the last 30 days it published 10 pieces, about 2.3 a week. These are counts of what was published, not a judgement of it.

Pieces tracked
39
2026-05-01 → 2026-09-07
Tickers covered
15
4 macro pieces besides
Directional statements
13
of 1,056 claims extracted
Last 30 days
10
about 2.3 pieces a week

Full-history audit result

SemiAnalysis is one of 3 sources (of 79 examined) that passed Quant Data's finance-influencer audit, which backtests every dated directional statement in a source's full archive against QQQ. Over 1,378 statements on 257 tickers from 2020-05-22 to 2026-07-07, the rule "first bullish statement, then hold" produced 57 trades with a 56% win rate and a median excess return of 9.3% over the benchmark. Method, caveats and the failures of other sources are on the audit page; the dated statements are pinned on price charts.

What kind of claims it makes

Claim typeCountShare
numbers on the record49547%
competitive notes14414%
risks13313%
theses979%
catalysts656%
falsification points485%
contrarian points353%
omissions flagged162%
directional statements131%
valuation points101%

Claim types are assigned during ingestion. "Omissions flagged" counts places where the reader noted a piece gave a view without a target, a timeframe or a falsification condition.

Tickers it writes about most

Each piece is tagged with one primary ticker; passing mentions are not counted.

Most recent pieces

DateTickerTitle
2026-09-07GOOGTPU Inference Externalization Full Steam Ahead - InferenceX
2026-09-01000660.KSKorea’s Trillion-Dollar Sovereign AI Investment
2026-08-30NVDAMost Neoclouds Suck At Security
2026-08-25NVDAOpenAI Jalapeño: Better Than Nvidia Blackwell
2026-08-24NVDAAgentX - InferenceXv3: Does CUDA Moat Hold up in Agentic Inferencing?
2026-08-21Are Open Models Catching Up?
2026-08-19Cerebras's Next Generation CS-4: Fast Just Got Faster
2026-08-17$12B of US ratepayers' money wasted on a modeling mistake and PJM wants to do it again
2026-08-16$12B of US ratepayers' money wasted on a modeling mistake and PJM wants to do it again
2026-08-10NVDAUltra-High Interactivity on NVIDIA GPUs? - TileRT InferenceX
2026-08-07GOOGLGemini is Cooked but GCP is Cooking
2026-08-07MSFTSpaceX 10GW in 2027 – Why It’s Real, Will Drive $500B ARR for SpaceX, and Why Microsoft Will Be the Largest Offtaker

Days with a written note

Counts are computed from Quant Data's ingestion ledger of SemiAnalysis's pieces accessed through the GeoScope aggregator, which licenses redistribution. Titles belong to the publisher. Educational only: not investment advice, not a recommendation.