Author

@GavinSBaker GavinSBaker

Institutional growth PM dissecting the AI compute stack with engineering-grade depth

Posts original, technically deep fundamental analysis of AI

trader score
+0.67
hit rate
44%
mean α
+0.52%
signals 14d
27

Grade = how their written analysis reads (A best). Trader score = how their last-20 timestamped calls performed vs SPY.

Their picks, scored

Across their last 18 scored bets: 44% hit rate, +0.52% mean alpha, trader score +0.67. Their last-14d mentions, direction-adjusted, have moved -2.9% since posting (mean over 2 mentions with price data).

Recent signals15receipts included
date (PT)tickerauthorsentwhat they saidsince thenreceipt
2026-08-27·@GavinSBaker·Strongly positive view of U.S. data-center development.·
2026-08-25·@GavinSBaker·Highlights that Jalapeño entered a third-party benchmark before TPUs.·
2026-08-25·@GavinSBaker·Compares model-size tradeoffs between interactivity and throughput and flags AFFN disaggregation as a variable.·
2026-08-25AMZN@GavinSBaker+0.20Jalapeño beats a comparable TPU but may trail disaggregated GPU/Trainium-plus-SRAM configurations.+2.1%
2026-08-24·@GavinSBaker·Argues frontier models have greater intelligence density and compute efficiency per token.·
2026-08-24·@GavinSBaker·Views open-source model share gains as positive broadly but negative for model companies.·
2026-08-24·@GavinSBaker·Expects Ultrafast AI capacity constraints to ease as wafer, server, and energization cycles progress.·
2026-08-24·@GavinSBaker·Dismisses concern over Fable 5 usage data as evidence of customer preference for ZDR.·
2026-08-24·@GavinSBaker·Predicts disaggregated inference will become the dominant AI-serving architecture.·
2026-08-24·@GavinSBaker·High-conviction view that enterprises will use a data-safe hybrid AI architecture.·
2026-08-23·@GavinSBaker·Expects broad adoption of zHBM or a similar memory architecture.·
2026-08-23·@GavinSBaker·Expects accelerator firms to adopt 3D DRAM or zHBM, with both memory types coexisting long term.·
2026-08-23·@GavinSBaker·Rapidly rising internal AI spending supports a thesis that token scale is becoming competitively necessary.·
2026-08-23·@GavinSBaker·Explains compute savings through the margin gap between frontier and open-source tokens.·
2026-08-23·@GavinSBaker·Says relative inference margins will determine the competitive outcome.·

Grade is our human read-worthiness rating; trader score is a rolling 20-bet hit-rate/alpha composite — different things, often disagreeing. “Since then” is direction-unaware in the table; the summary line above adjusts for which way they leaned.