Author

@akramsrazor akramsrazor

Deep AI-infrastructure bottleneck analyst turning macro frameworks into specific stock calls

Builds original fundamental theses around AI compute/power/m

trader score
-0.76
hit rate
55%
mean α
-0.35%
signals 14d
11

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 20 scored bets: 55% hit rate, -0.35% mean alpha, trader score -0.76. Their last-14d mentions, direction-adjusted, have moved +0.3% since posting (mean over 3 mentions with price data).

Recent signals11receipts included
date (PT)tickerauthorsentwhat they saidsince thenreceipt
2026-08-27·@akramsrazor·Satirical framework warns that concentrated AI compute and debt-financed replacement create systemic risk.·
2026-08-27GTLB@akramsrazor+0.30Author discloses an OKTA long and argues enterprise software complexity benefits OKTA, NOW and GTLB despite AI doubts.+0.1%
2026-08-27OKTA@akramsrazor+0.35Author discloses an OKTA long and argues enterprise software complexity benefits OKTA, NOW and GTLB despite AI doubts.-3.9%
2026-08-27NOW@akramsrazor+0.30Author discloses an OKTA long and argues enterprise software complexity benefits OKTA, NOW and GTLB despite AI doubts.+4.5%
2026-08-23·@akramsrazor·Satirically reframes AI bull and bear assumptions and anticipates a future compute glut.·
2026-08-23·@akramsrazor·Explains why token volume is an unreliable proxy for physical compute use or serving cost.·
2026-08-23·@akramsrazor·Argues useful intelligence is rapidly deflating in cost and mature agents may require far fewer LLM calls.·
2026-08-23·@akramsrazor·Argues DeepSeek-like models use fewer FLOPs per token and could drive major compute deflation.·
2026-08-20·@akramsrazor·Argues reported usage reflects ChatGPT distribution and says Codex had 5 million weekly users before changes.·
2026-08-18·@akramsrazor·Criticizes a comparison by noting the server market was about $10B quarterly and racks and data centers were much smaller.·
2026-08-18·@akramsrazor·Argues memory, not intelligence, limits agents and is central to future AI consumption revenue.·

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.