AI Drug Discovery Pipeline

Exiting the least AI-native holding in the book. Twist Bioscience makes synthetic DNA on a silicon platform — genuinely clever manufacturing and a real picks-and-shovels supplier to genomics — but it is an inputs vendor, not a machine-learning discovery platform, and it is the one position with zero return since entry and no momentum to wait on. With the ML Drug Discovery Platforms bucket 14 points under mandate, a flat 4.9% in a consumables supplier is the clearest source of funding for RXRX and CRSP. No thesis break here, simply the weakest claim on capital in a portfolio that has better uses for it.

Reducing redundant preclinical gene-editing risk. Intellia is negative versus our entry, printed -3.4% today, and its thesis — in-vivo CRISPR for ATTR amyloidosis and HAE — overlaps substantially with what we already own in BEAM while lacking the commercial validation CRSP now has. Holding three similar pre-revenue editing stories at 13% combined is diversification in name only; they share the same regulatory, delivery and financing risks. Cutting NTLA to 4.5% preserves a real option on its lead programs while freeing capital for the CRSP initiation, which buys the same modality with approved-product revenue behind it.

Taking partial profit on a +36.3% winner after a 60% surge. Schrodinger's physics-based computational chemistry platform is a core mandate holding and stays a core mandate holding — the thesis that structure-based simulation compresses discovery timelines is intact and the co-founded pipeline gives us free optionality. But the stock has run hard into flagged $37 resistance, GF Value calls it still overvalued after a -4.0% session, and president Karen Akinsanya sold $927,970 on 09-22. Trimming roughly two points into that strength harvests the move while leaving a 10% core position, and the proceeds fund RXRX where the same theme is priced at a fraction of the multiple.

Trimming a +30.8% winner that is the least disruptive name in the genomics sleeve. Illumina is the sequencing incumbent — its economics are instrument-and-consumable, it faces real competition from lower-cost platforms, and it is a supplier to the AI drug discovery stack rather than a participant in it. That is a fine holding but not where a disruptive-innovation mandate should carry 9% while the ML platform bucket runs 14 points light. Taking it to 7% keeps the read-depth exposure and monetizes strength into names with steeper adoption curves.

Tempus AI remains the anchor of the Clinical AI sleeve — its multimodal oncology data network and the Foundation/AstraZeneca-scale partnerships are exactly the disruption the mandate wants — and this is a trim on position size, not a change of thesis. Two reasons to take it down. At 14.2% it was the single largest weight in a book whose mandate caps this bucket at 25% in aggregate. And the last week produced an unusually dense cluster of insider selling: CEO-linked entities disposing of 250,000 shares on 09-24, the chief accounting officer $529,650 on 09-20, plus EVP and chief legal officer sales on 09-19. Individually these are preset 10b5-1 trades; arriving together after a +3.4% day they argue for taking size down rather than adding. Holding 12% keeps full core exposure.

CRISPR Therapeutics is the only gene-editing platform in this book that has crossed from preclinical promise into commercial reality — Casgevy is approved and reimbursed, and revenue is compounding +1041% YoY off that ramp. This is precisely the adoption inflection the mandate's genomics sleeve is built to own: the editing modality has cleared regulatory and payer risk, and the next decade is an expanding TAM across sickle cell, beta-thalassemia, and the in-vivo pipeline. Analysts target $87.56 against $54.23 today, a 61% gap, and today's -2.6% print is an entry rather than a warning. Funding it by rotating out of pre-revenue editing exposure (NTLA) upgrades the sleeve from optionality to a commercial leader without adding net genomics weight.

Recursion IS the ML drug discovery platform this mandate names first, and at 3.9% the book barely owns it. Two structural facts drive the add. First, the mandate's third named platform, EXAI, no longer exists as a tradeable security — Exscientia was acquired by Recursion and delisted in November 2024, its last real print 2024-11-19 — so the Exscientia thesis is now expressed inside RXRX, which absorbed the platform, its pipeline and its automated chemistry stack. Second, RXRX at $3.74 is priced as though the maps-and-models flywheel has failed, when what has actually happened is a normal capital-intensive buildout of the largest proprietary biological image dataset in existence plus the Exscientia merger integration. Recent CEO and director sales were 10b5-1 preset plans, not signal. Raising this to 9% moves the 14-point underweight in ML Drug Discovery Platforms toward the mandate while buying the flagship disruptor near its lows — the asymmetry we accept 50% drawdowns for.

AbCellera's AI-driven antibody discovery engine is the second pillar of the ML platform sleeve and the screen's headline -76% revenue decline is a known artifact, not deterioration: it is the roll-off of one-time Lilly COVID antibody royalties, which tells us nothing about platform throughput. What matters is that the company has moved from licensing discovery to owning clinical-stage assets, with wholly-owned programs now in the clinic and a GMP manufacturing facility online. Analysts carry $16.67 against $12.70. The position is already +16.6% for us and near its 52-week high of $13.18, so this is adding to a winner whose adoption curve is inflecting, funded from the overweight genomics sleeve.

Consolidating the book's pre-revenue editing exposure into the more differentiated platform. Base editing makes single-base changes without the double-strand break that conventional CRISPR requires, which is a genuinely distinct technical bet rather than an incremental improvement, and BEAM's ESCAPE conditioning and in-vivo liver programs are further along than NTLA's. Paired with the NTLA trim and the CRSP initiation, this concentrates editing risk into one commercial leader plus one differentiated platform instead of spreading it thinly across three similar preclinical stories.