Author · brief 2026-06-22

@institLPGP institLPGP

grade Ax.com/institLPGP ↗

Anon AI-hardware PM threading model-architecture shifts into memory/WFE/neocloud supply-demand calls

Builds and trades a concentrated, interlinked AI-infrastruct

trader score
-1.48
hit rate
35%
mean α
-1.56%
signals 14d
90

Grade = how their written analysis reads (A best). Trader score = how their last-20 timestamped calls performed vs SPY. · Analyst brief as of 2026-06-22.

Their picks, scored

Across their last 20 scored bets: 35% hit rate, -1.56% mean alpha, trader score -1.48. Their last-14d mentions, direction-adjusted, have moved -5.7% since posting (mean over 72 mentions with price data).

Long NBIS, CENX, ACMR while questioning AI incumbents

Author is most visibly long NBIS, CENX, ACMR, and still owns NVDA while arguing Nvidia needs an open-source/cloud strategy to protect GPU economics. Distinctive read is that Nebius is a strategic open-source AI infrastructure asset, CENX is a mispriced aluminum supply-cost curve play, and ACMR has a China memory/listing-arb catalyst. Late window added a fresh ACMR Hong Kong listing catalyst and remained strongly bullish NBIS after earlier trimming AI exposure for cash.

Themes5analyst read · 2026-06-22
◆Open-source AI cloud and GPU control
mixedconsistent64 signals
NBISNVDAMETAGOOGLAMZNSHOP
▲Aluminum supply and power-cost squeeze
bullintensifying8 signals
CENX
⚠ 100% of theme signals are CENX — flag pump risk
▲China memory equipment and listing arbitrage
bullintensifying9 signals
ACMR
⚠ 100% of theme signals are ACMR — flag pump risk
▼Regulatory capture and hyperscaler AI politics
bearfading25 signals
GOOGLAMZNMSFTORCL
◆AI hardware share and speculative pair trades
mixedconsistent18 signals
SPCXTSLAAMDGOOGTSEMMSFT
⚠ 50% of theme signals are MSFT — flag pump risk
Direction this week

Conviction is concentrated in NBIS, CENX, and ACMR, with explicit CENX adds and repeated NBIS long disclosures. He remains an NVDA holder but frames it as strategically vulnerable unless it secures cloud and open-source distribution. Pump-risk is highest in single-name themes CENX and ACMR, while the AI infrastructure theme is mixed because NBIS/NVDA upside is paired with warnings on Nvidia pricing power and hyperscaler capture.

Position disclosures7skin in the game
ACMR Cut some ACMR while raising cash due to uncertainty over AI path trimmed
NVDA Disclosed owning Nebius while saying NVDA has most to lose from frontier lab regulation unclear
NVDA Nvidia is second-biggest position, but model duopoly could threaten pricing power held
NBIS Owns NBIS and argues Nvidia should buy it for open-source strategy held
CENX Holds cash against CENX and sees near-double potential over 24 months held
NBIS Explicitly disclosed long NBIS held
NVDA Owns Nvidia as a hedge despite high correlation held
Best hypotheses5their highest-scoring claims in our index
▲Nvidia's software, CUDA, networking attach, and platform breadth make custom ASIC displacement harder than bears assume.
bullHIGH4 co-supporters
◆Some investors prefer trimming or rotating AMZN exposure after strength while waiting for cleaner entries or other AI wi
neutralHIGH5 co-supporters
▲Memory, HBM, and wafer expansion cycles should lift semiconductor equipment demand and benefit ASML.
bullMEDIUM3 co-supporters
▲Visible long disclosures from concentrated investors reinforce ACMR as a favored semiconductor long despite high-beta ri
bullMEDIUM2 co-supporters
▼Supply events, ETF outflows, and crowded longs could extend near-term downside despite the long-term thesis.
bearMEDIUM4 co-supporters
Recent signals15receipts included
date (PT)tickerauthorsentwhat they saidsince thenreceipt
2026-07-02MU@institLPGP-0.55Author argues Apple's China memory lobbying and MU donation signal political pressure to raise capacity and drop prices.—
2026-07-02AAPL@institLPGP+0.00Author argues Apple's China memory lobbying and MU donation signal political pressure to raise capacity and drop prices.—
2026-07-01·@institLPGP·Says DRAM should be fine for 12 months if GPU demand supports margins.·
2026-07-01AMD@institLPGP+0.20Thesis that capex persists as GPU rental/buying improves FCF and creates arbitrage.-4.3%
2026-07-01NVDA@institLPGP+0.20Thesis that capex persists as GPU rental/buying improves FCF and creates arbitrage.-1.4%
2026-07-01NVDA@institLPGP-0.40Discloses portfolio rotation: new AVGO/META, out of NVDA into DRAM, sold ACMR, cash 12%.-1.4%
2026-07-01DRAM@institLPGP+0.50Discloses portfolio rotation: new AVGO/META, out of NVDA into DRAM, sold ACMR, cash 12%.·
2026-07-01ACMR@institLPGP-0.20Discloses portfolio rotation: new AVGO/META, out of NVDA into DRAM, sold ACMR, cash 12%.-16.6%
2026-07-01META@institLPGP+0.50Discloses portfolio rotation: new AVGO/META, out of NVDA into DRAM, sold ACMR, cash 12%.-4.9%
2026-07-01AVGO@institLPGP+0.50Discloses portfolio rotation: new AVGO/META, out of NVDA into DRAM, sold ACMR, cash 12%.-2.4%
2026-07-01META@institLPGP+0.65Says he added 12% after META signal and views it as a big inflection.-4.9%
2026-07-01DRAM@institLPGP+0.65adding DRAM and META here — Author adds DRAM and META on GPU/cloud arbitrage thesis and compute bullishness.·
2026-07-01META@institLPGP+0.80adding DRAM and META here — Author adds DRAM and META on GPU/cloud arbitrage thesis and compute bullishness.-4.9%
2026-07-01META@institLPGP+0.35Strategic view that Meta pivots back to open models despite compute pricing concerns.-4.9%
2026-07-01META@institLPGP+0.75Author says it is very bullish Meta and points to open-source pivot strategy.-4.9%

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.