User: "GENB and ABCL are worth deep dives for sure to see where they're going." Built both, Massive-financials-backed, lint clean, projected.

• ABCL (2026-06-04-abcl-deep-dive.md, conviction medium): cash-backed OPTION not compounder — $0.50B cash + $0.94B book vs $1.75B mcap (EV ~$1.1-1.25B), ~$35M/q burn = 3+yr runway = real floor. BUT 3rd-most-extended in the lane (+77% 3m, RSI 67, at 52wk high) on AI-bio SENTIMENT — worst entry geometry. Own/hold; ADD on pullback ~$4.80-5.40, not at the high. Stop ~$4.20, cash-floor ~$3.50.
• GENB (2026-06-04-genb-deep-dive.md, conviction low): purest public generative-AI-protein bet (Flagship = public Lila-cousin). BETTER entry (below $16 IPO, neutral RSI, not extended) but WORSE balance sheet — $0.52B cash, ~$80M/q burn (2x ABCL), ~2yr runway → must partner/re-raise on a clock. EV ~$1.1B, EPS -$1.07, one quarter public. Speculative venture option; starter ~$13-15 defensible because geometry is clean, sized as venture, stop ~$11.
• Pair framing: barbell on the AI-protein-design layer — ABCL = wait-to-add on weakness (good balance sheet, bad entry); GENB = start on the below-IPO discount (good entry, runway clock). Neither core; both options on the same thesis.
• Portfolio tell surfaced: user's 3 lane holdings (TMO/TWST/ABCL) cluster — TWST+ABCL both already-run-winners tier; only TMO is a laggard. Un-owned optionality = Tier-1 laggards (MEDP/RGEN/IQV/CRL/DHR) + Tier-2 app-layer (SDGR/GENB). Adding more ABCL doubles down on the most-extended tier.