Thrilled to announce $5B in funding for @IsomorphicLabs to accelerate our mission to Solve All Disease. We continue to make rapid progress reimagining drug design from first principles with AI to radically cut the time & cost to find new medicines. Gradually, then all at once...
I received a lot of DMs and comments on this tweet. To be clear, AI doesn't guarantee drug approval or replace wet-lab validation and clinical trials. It's a force multiplier on the 2 biggest constraints: (1) low probability of success (~90% of drugs fail and ~10 of patients achieve remission of chronic disease), and (2) slow, manual operational processes. The question isn't "Will AI magically invent a cure tomorrow?" but "Given the current inefficiencies, can we afford not to use systems that could raise the quality and speed of attempts?" If you combine the intelligence of the smartest people who ever lived with everything humanity has written and the ability to run millions of simulations, it would be odd, with full-throated confidence, to claim that AI could not help find better drugs. The data so far says it already is.
The Biotocracy is up in arms about the “cure all diseases in 10 years” comment.
Future cure ≠ FDA Approval (2026)
Will the regulatory pathways and timelines look different in 2031, 2036? YES
Could there be a scientific breakthrough tomorrow, that the scientific consensus could not conceive of yesterday? YES
Just look at the recent DREADDs data for epilepsy that stunned.
e/acc
NEW from @FT: Johnson & Johnson is poised to buy privately held eye disease drug developer Ollin Biosciences for up to $3.4bn, the latest in a glut of bolt-on acquisitions in the biotechnology sector.
This is also true with biotechs clustering around bispecifics - the same way they cluster on the same targets.
Recent bispecifics have disappointed: J&J’s $1.25B IL-4/IL-31 failed its Phase 2b interim in atopic dermatitis, Genmab/AbbVie’s Epkinly missed overall survival in Phase 3 r/r DLBCL, and Compass’s DLL4/VEGF missed OS in biliary tract cancer (stock -50%).
Bispecifics lock you into a molecule with fixed ratios and linked PK/PD. Powered by CoFormForge (our AI engine), we select independent mAbs, optimize their relative doses, and co-formulate them into one product.
The approval committee of Anthropic's Life Sciences Verification Program (LSVP) bookmarking tweets of every biotech operator dunking on Dario’s “cures in 10 years” statement.
The Biotocracy is up in arms about the “cure all diseases in 10 years” comment.
Future cure ≠ FDA Approval (2026)
Will the regulatory pathways and timelines look different in 2031, 2036? YES
Could there be a scientific breakthrough tomorrow, that the scientific consensus could not conceive of yesterday? YES
Just look at the recent DREADDs data for epilepsy that stunned.
e/acc
"Now it also turns out the quality of the work [in China] is outstanding because guess who's doing it? A lot of Chinese scientists who got their PhDs in the United States and then we kicked out because we're….”
.@dawallach explains why China's rise in drug development could be a bigger disruption to the biopharma industry than AI:
"Over the past 5-10 years, China has both invested tremendously in early stage development, like discovery work, and pre-clinical biotech outsourced labor."
"In China, what they have tried to do is [all the work we do in America] at a quarter of the price. And because the big pharma companies are multinationals, increasingly they've been comfortable doing that work in China."
"Now it also turns out the quality of the work is outstanding because guess who's doing it? A lot of Chinese scientists who got their PhDs in the United States and then we kicked out because we're idiots."
"China is ascendant and now the lifeblood of big pharma, which is buying new drugs from little biotech companies. Now they are buying double-digit percentages of their new drugs from China."
I will be in Boston next week (11/10-11/14) for the Decoding Bio Investor Summit; and also for investor and partner meetings for my AI x Bio immunology startup, CoFormBio. Drop a DM if you’d like to meet.
I don't understand.
They used a bunch of open/closed-source tools to screen a tiny virtual library and are complaining they didn't generate a working drug?
Those are two entirely different design objectives. What does working even mean in this context?
I can't believe this is the state of discourse. Using computers to make medicines has been and will continue to be central to the art.
Small molecule tools are lagging behind biologics. They'll catch up. I think everyone knows this.
Non-story.
‘“That’s complete BS,” said Jeff Leiden, executive chairman of Vertex Pharmaceuticals. Vertex has used roughly 50 AI companies to comb through its data set of 10,000 drug compounds, Leiden said. Not one managed to generate a working drug.’ 👇🏼
wsj.com/tech/ai/silicon-vall…
Susan would be so happy to see the launch of the SWF foundation on YouTube. Thanks to all the creators for their support!! youtu.be/jJ7r6p9Qjes?is=jLRA… via @YouTube
Having waded through 163 pages of the draft COINS rule published by Treasury, my team and I can now respond with a simple graphic.
It will blind US companies and undermine our biopharma capabilities and national security, exactly opposite to what the policy is aiming to achieve. It does nothing to actually block a Chinese biotech from commercializing a drug in the US, with a European partner, European investors, or on its own. We value medicines highly and will now be paying others for them. Why would we do this to ourselves? We shouldn’t.
But you can’t submit a comic to Treasury so we wrote out a detailed comment that both lays out how the draft rule would backfire on American biotech and what revisions would allow Treasury to better achieve their stated national security objectives.
Have a look. Biotech investors and executives are welcome to sign onto our comment but the main purpose of publishing it now is to drive discussion that helps everyone sharpen their thinking and informs their own comment to Treasury, due Dec 7th. Anyone is welcome to crib from this. No need to ask.
We wrote it after considerable input from legal experts and peers. Since coins draft rule is itself unclear on many points, reasonable people can disagree as to its intent and meaning, so some of our comments address broader points meant to encompass a range of possibilities.
If you read what Treasury wrote, you might get the sense that they know this will backfire on America and are looking for support to revise their language to avert harm. So give this thought and submit your own analysis. We provide links in the article to the submission page.
And thanks to all who have already signed on.
@NPLB_orgracap.com/rapport/comment-fr…
While the world focuses on election results, 33M people across the US, EU5, Canada & Japan live with moderate-to-severe immune-mediated disease, and >90% (31M) still don’t achieve full remission or access advanced therapy. CoFormBio is building therapies for the >90% left behind.
As someone who has helped companies pioneering the use of AI for drug discovery, this is BS. Sure, AI is a tool that we all should deploy to the best of our abilities, but the idea that AI is going to unveil new medicines at a stunning rate the way it will upend cinema, video games, financial services, advertising, gambling, etc., is a bit absurd. New medicine discovery will be one of the laggards in terms of AI revolution. Healthcare provision, diagnostics, and paperwork streamlining will all move ahead faster.
Biology will continue to eat everything, AI included.