$ARM

Does he own it?

Strong first-person evidence: he explicitly stated going long at ~$139 in late March, then repeatedly referenced "since I took positions / went long" and tracked his personal ARM return (e.g. +41.5%) as one of his 100%+ YTD longs. Undercut only by a June 9 disclosure listing financial interest in NBIS/TSEM/AAOI but not ARM, hinting he may have since trimmed or exited.

His thesis

Bullish on a structural shift in AI compute from training toward inference, where CPUs (largely ARM-based: AWS Graviton, Google Axion, Microsoft Cobalt) reclaim share as GPU:CPU ratios rise and lightweight/edge/local inference grows. Centers the thesis on ARM's new "AGI CPU" with hyperscaler/OpenAI/Meta as customers and a ~$15B incremental annual revenue target (~5x current revenue), arguing the entry valuation underpriced that optionality.

How his view evolved

Started as a high-conviction buy thesis at ~$135-139 in late March, then shifted to repeated victory-lap tracking as it tripled. By early June he frames ARM as having "run quite a bit" and omits it from his active financial-interest list, implying a likely trim or exit.

Key points

• Explicitly went long at ~$139; rode it to a ~3x and tracked it as a personal holding
• Core driver: inference-era CPU demand and the new ARM AI CPU with ~$15B revenue target
• June 9 disclosure omits ARM from current financial interests, suggesting he may have exited

Derived from a holistic read of his public posts — paraphrased, never quoted. See his actual posts about $ARM on X ↗