
Chief Mid-Curver | Prev: @Shitcoinmaxi_ | Everything posted is NFA | DYOR Founded @SmResearchLab
Joined January 2025
- Tweets3.5K
- Following884
- Followers2.9K
- Likes3.5K
OpenAI just launched MedChemBench, a test of how well AI handles real drug chemistry: predicting potency, toxicity and absorption, and picking which molecules to make next.
Its best model scores under 50% (48%).
GPT-Rosalind, OpenAI's dedicated life-science model, scored 27.5% in September.
General AI still struggles with drug design.
That's the gap $IAM is built for. Enchant isn't a chatbot taught chemistry. It's trained on Iambic's own lab data, 500+ new compounds a week, and the platform has already produced a drug.
Seems like the frontier labs are racing into drug discovery.
A $0.7B company just put $GOOGL name on a chart. And Google lost.
$IAM's investor deck tests its AI, Enchant, against Google DeepMind's drug AI, TxGemma.
The questions every drug lives or dies on:
→ Will it get into cells?
→ Will cells pump it back out?
→ Will it hit its target?
→ Will it work in cancer cells?
→ Will it dissolve?
Enchant won all five. On three, Google's model barely registered.
Why? Google started with a general AI and taught it about drugs. Iambic built Enchant for drugs from day one, on its own lab data, and makes 500+ new compounds a week to keep it learning.
Alphabet's drug company, Isomorphic, is valued at ~$50B.
Iambic: ~$0.7B.
(Iambic's own 2025 testing.)
NFA, DYOR
Genentech paid $850M upfront for Regor's CDK cancer drugs.
That's more than all of $IAM is worth today (~$650m), and ~$455M of that is cash.
For less than that one deal, you get 15 programs. One of them is IAM-C1, a CDK drug in the same class, and in Iambic's mouse tests it matched or beat Regor's lead drug, the one Genentech paid up for.
It also takes on Pfizer. Pfizer needs two drugs to do what IAM-C1 does with one:
Tumour growth blocked
→ IAM-C1: 95%
→ Pfizer's two-drug combo: 95%
→ Ibrance: 60%
White blood cell drop (the side effect that holds these drugs back)
→ IAM-C1: −7.5%
→ Pfizer combo: −43%
→ Ibrance: −69%
Same result as Pfizer's two drugs. One molecule. A fraction of the side effect.
One CDK portfolio: $850M.
All of Iambic, 15 programs plus ~$455M cash: ~$700M.
IAM-C1 is due to file for human trials this quarter.
(Preclinical, cross-study.)
NFA, DYOR
4
759
The future of pharma is AI-designed drugs. $IAM's investor deck shows why.
Hit-to-lead failure rate
Industry: 25%
Iambic: 7%, about 1 program in 15
Preclinical failure rate
Industry: 40%
Iambic: 0%. None of its 3 preclinical programs has failed
Most of pharma's money gets burned on drugs that die along the way. Cut the failures and you change the economics of the whole industry.
What's doing it, per the same deck:
→ 500+ new compounds made every week
→ 300+ in-house assays
→ 12M+ model predictions a month
→ Models fine-tuned every week
Takeda's CSO, quoted on the same slide: Iambic could help "de-risk candidate selection, improve probability of success".
There is a reason why $NVDA sold their whole $RXRX stake and doubled down on Iambic.
NFA, DYOR
$IAM
Found a lot of interesting information in here, that we haven't seen before.
Will be posting today.
Looks like the dots were right.
Last week I said $IAM was quietly building towards obesity. Their November investor deck just put it in a footnote.
What I had:
→ Old job ads showed an in-house GPCR team. Ozempic, Wegovy and Mounjaro all work through GPCRs
→ The Q3 release says Iambic runs its own programs in "oncology, metabolic and endocrine diseases"
→ Enchant v3 is trained on peptide data
What the deck adds: three undisclosed programs (I&I/endocrinology, oncology, metabolic disease), sized at >$75B.
The footnote shows what that number is built on: 2025 sales of Mounjaro, Zepbound, Wegovy, Ozempic, Trulicity and Tepezza.
This isn't just a cancer company. Iambic makes small molecules, the kind you take as pills, and it's sizing its next wave against the GLP-1 market. Obesity pills look like where it's heading.
No obesity drug has been named yet. But the direction is now in their own deck.
NFA, DYOR
$IAM float maths, from the 10-Q filed today:
48.9M shares outstanding.
Only ~10.1M can trade, about $143M of stock.
The other 38.9M (79%) are locked up until 12 April.
The tradeable float is worth less than a third of Iambic's cash (~$455M).
Now add the index bid:
→ $IAM is on FTSE Russell's preliminary additions list. At ~$700M it lands in the Russell 2000
→ Russell only counts shares that aren't locked up, so its float is ~10M
→ Nasdaq estimates Russell 2000 trackers hold ~10% of a small cap's float
→ That's ~1M shares Russell 2000 funds need to own by the close on 11 Dec, about two days of normal volume
And you're in the same float as the funds. A fund building a position can't hide in a float this small.
*cough cough* Druckenmiller.
NFA, DYOR
$IAM filed its 10-Q today. More receipts for the bull case:
→ AbbVie signed two deals, not one: a drug collaboration plus an "AI-Discovery Agreement" that licenses AbbVie its own instances of Iambic's platform. $47M upfront, paid in October. That one payment is bigger than Iambic's entire Q3 operating costs ($41.5M).
→ Takeda is ramping. $7.8M of revenue in Q3 alone, more than the whole first half ($6.2M). It also paid its first milestone within 8 months of signing.
→ Q3 revenue was $11.4M, up 142% on last year.
→ ~$101M of signed partner work is still to be booked as revenue, before a single milestone. AbbVie's milestones alone go up to $670M.
→ Cash runway reaffirmed into Q4 2028.
All of that for ~$240M after cash.
NFA, DYOR
$IAM filed its 10-Q today. More receipts for the bull case:
→ AbbVie signed two deals, not one: a drug collaboration plus an "AI-Discovery Agreement" that licenses AbbVie its own instances of Iambic's platform. $47M upfront, paid in October. That one payment is bigger than Iambic's entire Q3 operating costs ($41.5M).
→ Takeda is ramping. $7.8M of revenue in Q3 alone, more than the whole first half ($6.2M). It also paid its first milestone within 8 months of signing.
→ Q3 revenue was $11.4M, up 142% on last year.
→ ~$101M of signed partner work is still to be booked as revenue, before a single milestone. AbbVie's milestones alone go up to $670M.
→ Cash runway reaffirmed into Q4 2028.
All of that for ~$240M after cash.
NFA, DYOR
Another comp that makes $IAM look even cheaper.
Jan 2025: Menarini licensed Insilico's AI-made KIF18A cancer drug before it had reached a single patient. $20M upfront, $550M+ headline value. It's in Phase 1 now.
Iambic is worth ~$0.7B, ~$240M after cash. For that you get:
→ IAM217: its own KIF18A drug. FDA-cleared, and ~90% shrinkage of breast tumours inside the brain in mice
→ IAM1363: a HER2 pill already in 100+ patients
→ IAM-C1: a CDK2/4 drug, IND due this quarter
→ NeuralPLexer and Enchant, the two AI models that made all three. Peptides are now one of Enchant v3's data types
The headline value of that one rival deal is more than 2x Iambic's value after cash. And when Menarini signed, that drug was at an earlier stage than IAM217 is today.
Add it to the pile: Isomorphic at ~$50B with no drug in human trials. Genentech paying $850M upfront for Regor's CDK drugs.
Whatever comp you look at, whether it's what rivals with less to show are valued at or what pharma pays for similar drugs, Iambic is cheap.
NFA, DYOR
6
875
$IAM's brain cancer drug just went head to head with a rival that's been in human trials since 2023.
Breast cancer growing inside the brain (pre-human trials), same dose:
→ Iambic's IAM217: ~90% shrinkage
→ Volastra's VLS-1488: kept growing
Outside the brain, both worked. The difference is that IAM217 gets in.
Volastra is private and has never disclosed a valuation. Investors including Eli Lilly have put $130M+ into it.
Iambic's market value is ~$0.7B, and that covers IAM217 (FDA-cleared for Phase 1/2), a HER2 pill in 100+ patients and a third drug heading for the clinic.


















