The Filing

10-Q/A (Feb 26, 2026) is a nothing-burger. Amendment No. 1 corrects typographical errors in SOX certification exhibits — wrong form number referenced. "No other changes have been made to the Form 10-Q." Zero new financial data.

The original Q1 FY26 10-Q (filed Feb 2) is where the actual information lives. That's what this memo covers.

The Business

Twist Bioscience makes synthetic DNA using semiconductor-based manufacturing. The bull case is that AI drug discovery is creating a new market for their products — instead of just selling DNA strands, they're selling characterization DATA. Customers send in targets, Twist synthesizes, tests, and returns results. $50/fragment, $100/clonal gene, $200/antibody, $300-400/characterization data.

The numbers are real. >$25M in AI discovery orders booked in FY2025 in a market that "effectively didn't exist in 2024." Q1 FY26 revenue grew 27% YoY in their DSPS segment, driven primarily by AI customers. Three customer types: big pharma building LLMs, big tech "dry labs" outsourcing all wet lab work, well-funded biotech. Management claims customers are placing repeat orders.

Gross margins improved to 52% (+4pts YoY). EBITDA breakeven committed for Q4 FY26 (July-Sept). Full year guidance $435-440M (+16%). Cash position $197.9M with runway past profitability target. Investor Day announced as near-term catalyst.

Good company. Possibly great company.

The Problem: None of This Is Edge

$2.8B market cap. 12 analysts. Every one of them has written the AI drug discovery story. Forward P/E -24x (still losing money). Consensus target $52 (13% upside). This is as efficiently priced as a mid-cap biotech tools company gets.

Factor decomposition kills it:

• Beta: 2.29 (you're buying 2.3x SPY exposure)
• Idiosyncratic variance: 58.7% (below 75% threshold)
• ≈41% of your return variance comes from market/sector, not the company

TWST is up +24% 1Y. XBI (biotech sector) is up +48%. The stock is underperforming its own beta. You'd have been better off buying XBI.

Forward alpha calculation:

Consensus target $52, current $45.89, T=1yr, r_f=4.3%
Excess return = 13.3% - 4.3% = 9.0%
Edge% = 0% (no informational advantage on any factor)
α = 9.0% × Conviction × 0% = 0%

Even granting full idio edge (58.7%), α = 3.7%. On a 2.29 beta name with 27% short interest, that's a terrible risk/reward ratio.

Insiders Are Telling You Something (by Not Telling You Anything)

Pulled every Form 4 from Dec 2025 through Feb 2026. The pattern:

Selling: President/COO Finn selling every 2-3 weeks ($434K total). CFO Laponis sold 3x in 5 weeks ($567K). SVP HR Green dumped $2.07M in a single trade on Feb 17 — 24% of her position, 220x her prior sale pace. Director Chess sold 2,184 shares, leaving his smallest holding among all insiders. CLO Cho sold a small amount in Dec.

Buying: Zero. Not a single open market purchase (Code P) in three months.

All sales executed through same attorney-in-fact (Kendra Fox), all preceded by Rule 144 filings. This is textbook 10b5-1 plan execution. The plans were likely set when TWST was $30-35, with trigger prices at $40, $45, $50, $52.

The selling itself isn't alarming — stock doubled from its 52-week low, insiders monetizing into strength is normal. They retain large positions (Finn: 284K shares, Green: 126K, Laponis: 130K).

But the ABSENCE of buying is the signal. If management truly believed AI drug discovery would take this materially higher, you'd expect at least one insider adding at current levels. Nobody did. The people with the most information about the EBITDA breakeven trajectory and AI order pipeline are systematically reducing, not increasing, their exposure.

Green's $2M sale is the loudest voice. SVP HR doesn't usually make the biggest insider trade of the quarter. Could be personal (divorce, house, estate planning). Could be the 10b5-1 plan hitting a target. But it happened at $52.40 — right at the analyst consensus target. She sold INTO the target, not through it.

Options: Market Has No Opinion

Total OI across all expirations: ≈7,500 contracts. Daily volume: ≈300. The Greeks from yfinance are nonsensical (showing 0.00 delta on OTM options). This market is too illiquid to derive implied probabilities.

What we can see: P/C ratio 0.21 (bullish), $50 call strike is the OI magnet, IV rank 106-124% (above 52-week range), April expiry has the most OI (positioned around May 4 earnings).

The absence of institutional options activity is itself informative — big money isn't expressing views here through derivatives.

Risks the Market Might Be Underpricing

1. Customer concentration. One large NGS customer distorts growth. Excluding them, NGS grew 18% not 27%. Top 10 = 36% of NGS revenue. ≈75% of diagnostics is OEM (dependent on partner execution). Management admits they "need new partners for 2027+."

2. SG&A acceleration. +24% YoY ($13.6M increase). Personnel costs +$7.7M, outside services +$5.4M. Investing ahead of revenue. If AI orders plateau, this spend becomes a problem fast.

3. Litigation overhang. Scorpion Capital short report class action (Peters v. TWST) is in discovery phase. No accrual recorded. Discovery could surface information or force a cash settlement.

4. Short interest: 27.3%, 8.7 days to cover. This cuts both ways. Squeeze potential on any positive catalyst. But 27% of float betting against you is a lot of informed money.

Verdict

Good company, consensus story, zero forward alpha for anyone without proprietary insight into: (a) AI drug discovery order pipeline magnitude, (b) customer concentration identity and stability, or (c) EBITDA breakeven trajectory internals.

The 58.7% idio variance means you're paying for a 2.3x levered biotech sector bet wrapped in a company-specific narrative. If you want biotech exposure, XBI is cheaper and simpler. If you want AI drug discovery exposure specifically, you need a view on whether this market is $2B (management's SAM) or $500M (bear case), and 12 analysts are already fighting over that question.

Insiders are selling into the consensus target. Nobody is buying. The options market is too thin to express a view. Pass.