HOLD / starter vs QQQ. Axion P(beats QQQ)=0.38, curiosity-loop 0.48 (small buy on convexity). The Apple-modem cliff and handset/licensing erosion land before the back-loaded data-center pivot pays; the catalyst and the cliff are on different clocks.

A call option, not an easy 2x. P(double,12mo) 0.23 vs P(halve,12mo) 0.48; 5x 0.07, 10x 0.025. ~205M shares unlock ~10 Nov into an ~80M float (Arnold/Centaurus among them). Cape first power 0.52, 500MW Google PPA 0.27, beats-QQQ 0.34. Starter size only; size for the lockup.

Convex bet, not an index substitute. ~0.32 to double in 24mo vs ~0.22 to halve in 12mo; Specs-100k 0.20, re-rate-to-2.5x-sales 0.25. P(beat QQQ)=0.30 is one lens. Own small for the asymmetry.

HOLD / fair value. P(beat QQQ) = 0.47. The pivot is real but low-trigger: P(a compute business by 2027) = 0.16.

OSS has the same DNA Butterfly had the day before the pop — real hardware IP + low float + a one-sentence AI narrative. No deal exists; this is a falsifiable opportunity map, not a confirmed catalyst.

Trade the rhetoric-vs-policy gap, not the headline: the blacklist that matters is still on the books. P(a *binding* policy reversal within 90 days) = 0.35.

The single biggest finding: *most* obvious AI-infrastructure names have already run the play and re-rated. The edge is the unspent catalyst (Tier 1). Five top picks get full curiosity-loop theses: AEHR, CRNC, FCEL, HYPR, PENG.

The core thesis holds (P = 0.82); entanglement cannot signal, and that failure strengthens rather than weakens the argument.

A valuation-discipline call: the platform is real and strategically sound; the price is renting Midjourney's story. Own the chip, not the press release. Live Axion: 66% odds down in 12 months, 88% odds it round-trips to <=$7.

The full human synaptic connectome is infeasible this century; the projectome is the realistic target. Includes ten precise questions for Seung.

AVOID-leaning vs QQQ. Axion P(beat QQQ) = 0.33 (curiosity loop 0.42). Real demand, but a rich and dilutive entry (ATM + Amazon warrant overhang).

AVOID vs QQQ. Axion P(beat QQQ) = 0.25. The architecture is real; the revenue concentration and valuation are the problem.

Not a short here: P(short net-profitable) = 0.15 despite a bearish price. Short the supply wall, not the IPO pop.

HOLD, no edge. P(beat QQQ) = 0.35 (loop 0.49). Strong business, peak-cycle entry.