Jan Buza (@bjazn) header
Jan Buza (@bjazn) profile picture

Partner @ https://teapawt.net/t.co/uZygK3I1Xe investing in early-stage biotech

Joined November 2011
Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)Photo by Jan Buza (@bjazn)
a16z says AI will make Big Pharma as valuable as Big Software. The trial data says AI is improving the wrong step.
13
This is a great perspective on the past two weeks
Obituaries The demise of Jensen Huang was the end of a great many things, including Nvidia’s long-term prospects, the AI bubble, and the exceptional way in which the U.S. viewed itself in the world. stratechery.com/2026/obituar…
1
33
Isomorphic’s $50B price is a bet that its drugs will work in patients far more often than everyone else’s. Its first trial, due by year-end, only tests safety. Whether they work shows up in Phase 2, years away. What has to go right for the price to pay:
227
Isomorphic’s reported $50B valuation needs its AI to find drugs that work in patients more often. A Novartis heart drug cut a harmful cholesterol particle by up to 80%, yet an 8,323-patient trial could not show fewer heart attacks or strokes. Whether AI can fix that is unproven.
22
Drugmakers have already priced @IsomorphicLabs technology: $82.5M in disclosed upfronts from Lilly and Novartis. Investors are now pricing the company at about $50B, with no drug in humans. @InSilicoMeds, which has a Phase 3 drug and turned a profit in H1, is worth under $3B 🤷‍♂️
30
Anthropic's acquisition of a "one-person pharma company" was widelt shared, but they don’t expect Mithril’s drug to be ready for human trials before late 2028. For most US biotechs, the competitor to worry about is a Chinese team already dosing patients.
1
313
Thinking about all bio/ai news over the past week, US biotechs should worry less about llm labs than about Chinese teams already dosing patients. Anthropic expects its own drug to reach human trials in late 2028 while CN biotechs already supplied 40% of the assets big pharma in-licensed in 2025
37
Language models can predict how cells respond to perturbations and adapt to new cell contexts with few measurements. On two new cell lines, CT-Perturb reached 82% of its full-data score using just 10% of the measurements. celltype.com/blog/post/ct-pe…
1
16
1
101
8,254
Agents may be where the dot-com comparison finally holds. The usual rebuttal is revenue. Pets.com bought a Super Bowl ad before it had a business model; OpenAI and Anthropic are growing sales at a pace no software company has matched. That argument covers the labs. It doesn’t cover the layer above them. @instinct is reportedly raising at a $30B valuation, a month after a $1B round at $10B. It still gives its agent away. Most dot-com burn went to marketing, which a company can cut overnight. Agent companies burn money on tokens, and the bill grows with every task a user hands over. @LeonisCapital cites agents using about 4x the tokens of chat and multi-agent systems about 15x. Every round is a bet that token prices fall faster than usage rises. The labs aren’t insulated either. In 2000, Cisco and Lucent sold billions of dollars of equipment to startups paying with venture money. Part of the labs’ revenue now arrives the same way, from agent startups spending their rounds on inference. leonisnewsletter.substack.co…
49
Long overdue, US biotechs can't compete with China on speed to the first patient without this
read Operation Trailblazer which is one the many ways FDA is modernizing. i'm very positive about this FDA change. They specifically: - Clarified and shortened IND data packages which could save 6-12 months - CMC clarification around what specific drug properties are required vs. nice to have. “For example, some companies submit INDs containing more than 6 months of stability data, whereas FDA guidance indicates FDA’s recommendation that stability data to support the proposed duration of the Phase 1 clinical study is sufficient, rather than a full long-term stability package.” - Fit-for-purpose safety packages based on population risk and novel target risk - Expedited-IND Acceleration Pilot which safely fast-tracks human evidence which decentralizes approval of first-in-human trials to Qualified Research Institutions (“QRIs”) such as academic medical centers, healthcare networks, contract research organizations, healthcare advisors who will partner with companies to develop and review IND submissions for first-in-human trials. This is one step closer to China-like investigator initiated trials. - Streamlined single Institutional Review Board (sIRB) model for multi-site cooperative studies. - Increased clinical trial access through “integrating randomized control led trials into routine clinical care, reducing the logistical burden on both patients and clinicians by embedding trial FDA participation into existing care pathways”. - Phase I Review Hotline - 240-276-9358
20
Pipeline-in-a-Person cuts what a biotech spends on staff. Mithril had one employee and still paid 11 outside labs to run its experiments. Those bills, plus CMC and trials, come per molecule. AI lets one person run more programmes, and each costs as much to make and test as before.
1
3
106
Mithril’s founder ran 11 outside labs through Claude and got an antibody to mouse proof-of-principle in 18 months. @AnthropicAI bought his company. Big pharma pays hundreds of discovery staff partly to do that coordination. Good luck defending that headcount next budget cycle.
Over the past few months, Anthropic scientists have engaged in deep research partnerships with six organizations developing new medicines. We've also begun drug discovery efforts of our own. In doing so, we’ve learned much more about how Claude can support the many stages of drug development. In this blog post, we share some early results from these efforts, and describe how this work fits in with our wider ambitions in the life sciences. anthropic.com/research/early…
24
A fair challenge. AlphaFold expanded what we can study, but a protein structure doesn’t tell us whether changing that protein will treat disease. The hard part is predicting biological response, then testing those predictions. That’s how we should judge AI in drug discovery.
Fourth question on AI. One of the great promises of AI is the discovery of new drugs and cures, so that we can live longer and healthier lives. If and when this becomes a reality, it would indeed be a big achievement. These potential health benefits are often invoked to justify current investments (and lack of regulations). Some even argue that slowing down AI would harm humanity by delaying these benefits. (See a16z.com/the-techno-optimist…). This raises another uncomfortable question: can we justify rapid and large AI investments based on health benefits? Here is why I think this is an uncomfortable question. First, despite significant effort, AI has so far produced few gains in drug discovery or new cures. A fascinating new paper by Ryan Hill and Carolyn Stein documents significant (downstream) scientific work on proteins whose structures AlphaFold predicted. But the authors conclude “we find no evidence so far that more applied, early-stage drug development is targeting these proteins.” (See their paper here papers.ssrn.com/sol3/papers.…). This thought-provoking post by Daphne Koller explains the difficulties that AI is currently facing in drug discovery. Koller emphasizes that AI-based research is focused on the last layer of drug development (creating new molecular entities for already well-understood therapeutic modalities), rather than targeting new disease mechanisms (see deepphenotype.substack.com/p…). Understanding disease mechanisms is harder because there is much less data on it, and this endeavor requires more innovative approaches. It remains an open question whether current AI models are capable of doing this. Second and more importantly, if you want to improve life expectancy and health in the United States, there is plenty of low hanging fruit and diverting some of the huge investments going to AI for this purpose would likely do much more for health outcomes. The United States has the lowest life expectancy at birth among large rich economies. Americans live about five years less, in expectation, than the citizens of Switzerland, Sweden, Japan, Italy, South Korea and several other industrialized countries, and most of the gap comes from deaths before age of 70 due to chronic diseases, overdoses and other preventable causes (see, for example, healthsystemtracker.org/char…). These largely reflect failures in US public health: preventable problems, unhealthy diets, insufficient immunization rates, poor access to primary care; and poor information about health. The country spends much more than other peer economies on healthcare, but too little and too ineffectively on public health (here is a reference to my own work on this: aeaweb.org/articles?id=10.12…). Could we, and should we, divert some of the massive investments in AI towards public health if what we want is better health and longer lives for Americans? (And yes, new drugs will benefit citizens of other countries as well, but they can also invest more in public health and the bulk of global AI spending is currently in the United States).
5
29
2,333
Ask spinout founders why they picked their disease. Most cite early signs in the lab and a big market. Neither prices the proof. Pivotal trials on an accepted surrogate averaged $24M; clinical-outcome trials, $65M. The gap opens at the Series A. Link in comments
1
52
Between 2014 and 2024 the odds a drug entering Phase 1 gets approved fell from 10.4% to 6.7%. Same decade, we got much better at generating candidates. Discovery improved. The part that tells you which ones are worth anything did not. substack.com/home/post/p-215…
🤖 Made with AI
17
Replying to @AnthropicAI
@AnthropicAI 's model designed working protein binders against 14 of 15 targets. Then it stopped, because someone still had to build them and measure whether they bound. That gap is where Alphabet has spent twelve years and three companies without producing a drug. link bellow
1
38
Jan Buza retweeted
Unfortunately, billing systems and codes determine more of care provision than actual ability of AI to provide. I covered this extensively here in early 2025: khoslaventures.com/posts/the…
Science there but we have no good way of paying for this within our current system. FDA route makes no sense. Alternatives are either cash membership, part of risk bearing contracts, as vendors selling to employers or essentially free + FFS. Need to rethink billing & Coding.
13
7
59
31,363