Life sciences x technology x markets per aspera ad astra
San Francisco, CA
Joined July 2014
- Tweets1.5K
- Following1.2K
- Followers5.7K
- Likes15.2K
David Li retweeted
Would be interesting to apply the new b-school grad search fund model to newly minted MD/PhDs starting AI-native biotechs.
One of the hottest career paths right now for @HarvardHBS & @StanfordGSB students is to look for small businesses to buy and run on their own.
After college, they get some small amount of funding to cover their living expenses that then expands once they find a business to buy (HVAC, niche service provider, etc.) so they can actually buy it.
This wasn't really possible in biotech until this year. The act of finding assets, running trials, running a regulatory function has collapsed.
Could go something like ~$500k-$1M until you get an asset and then ~$10M to scale up, ~$5M for the asset & ~$5M to push it forward like next step of a trial (likely abroad), token costs. Ofc needs to fit a specific profile like rare diseases.
Have also talked to former hired biotech CEOs who are excited to be founders with more ownership and eager to re-invent the pharma model, have insight into how'd they capitalize on pharma relationships but acquire/advance assets much differently than they've done in the past.
@Curie_Bio, @FlagshipPioneer & others do some flavor of this. But haven't seen anyone target the biotech-equivalent recent b-school grad founder persona that matched PhD & MDs.
h/t to @davidycli for insights & ideas on this
David Li retweeted
THIS is how AI will (and is) making drug discovery more efficient. Simple stuff, just faster and more comprehensively.
** a vision for the future AI native workflow for drug discovery R&D**
scientist assesses emerging targets in a particular field of cancer, with assist of literature and competitive intel natural language co-pilot
uses expertise to decide target X is interesting to try to drug
queries patient standard of care competitive bar using natural language co-pilot
sets the target product profile of the molecule (ie "what" the molecule will do for "whom" who are at which stage of their patient journey)
co-pilot spits out target ranges for molecule profile in order for it to *likely* achieve target profile in clinic (potency, selectivity, oral bioavailability, solubility, time on target, etc.)
scientist explores different modality approaches to drug the target and achieve desired molecule profile with co-pilot, ie determining feasibility of small mlc inhibitor vs degrader vs peptide vs drugging upstream target with an Ab
for each modality, use co-pilot to call foundation models for protein folding / ligand interaction to determine potential binding pocket, protein protein interactions, and more
scientist weighs the feasibility of each modality and decides to go with small molecule inhibitor
scientist asks co-pilot for any existing patents / literature that could provide starting hits to the target, as well as areas that could be jumping off points for establishing freedom to operate (FTO)
scientist selects a few reasonable starting hits to begin with, asking co-pilot to synthesize and characterize
co-pilot automatically engages a CRO to synthesize the molecule, check quality, and run initial characterization assays
co-pilot automatically logs the results in a system of record and notifies scientist
scientist likes the selectivity profile of the hits but asks co-pilot to suggest updates that would improve potency at least 5x but retain significant solubility
co-pilot makes suggestions, and iterates with scientist two times before sending another round to CRO
updated molecules pass automatic ADME and tox modeling gates presented by co-pilot
scientist selects resulting molecule to enter mouse model testing, co-pilot suggests appropriate preclinical models for the cancer in question, as well as positive and negative control arms
another round of testing later, scientist asks co-pilot to start writing up IND-enabling GLP testing docs
finally, scientist asks co-pilot to draft a IND filing package for regulatory authorities including a first in human (FIH) clinical development plan based on preclinical testing results
********************************
how far away is this vision?
David Li retweeted
moolenaar's proposal would likely have the opposite of the intended effect. if u ban US pharma from acquiring chinese IP, what is likely to happen is EU/Japanese pharma will get that IP, and u hand them a structural advantage. if u somehow convince them to also ban china deals, then u force chinese biotech to forward integrate and compete directly sooner. hengrui and at least 10 other biotechs already have the resources to do so. then let's say u ban those companies from the US market, then u just have a replay of the EV story, chinese pharma will just take over the rest of the world.
the proposal also tries to scare people by drawing the analogy to chips and rare earths. there is no such analogy. licensing deals are for IP, and almost always include the right to manufacture wherever you want. so there is no supply chain risk. in fact, in drug world, the vast majority of the value of drug IP accrues to whoever develops and commercializes a drug, not who invented it. so right now western pharma is the winner from this relationship.
so if the Q is well how do we not lose to china longterm? the answer is we must compete. america is great, we should lean into our natural advantages. continue to lead in breakthru science, maintain fda's position as the preeminent regulator, apply our cutting edge technology, and entice the world's best with the most pro-innovation commercial market. these leads are larger than many realize. we should mitigate our weaknesses, most importantly the bureaucracy that comes from being around longer than others that slows us down.
Could we be nearing the end of the China biotech deal wave? At least one prominent GOP lawmaker is pushing for the government to restrict them.
endpoints.news/gop-lawmaker-…
David Li retweeted
I am going to outline the one (and only) strategy by which America can beat China.
While I personally think the two great countries can and should learn to cooperate and co-exist, the governing midwits of our current age don't think that way, so I'm going to lay out the battle plan.
Beware, China is probably reading this so you get one chance.
First, realize that China outmatches the US in almost all aspects of relevant national power today, both in main KPIs ("number of factories") and key drivers ("ability to build factories"). If you don't think so, you're too far behind the real story and should just save time and close the thread now. You're ngmi.
For everyone else:
The reason China is ahead of the US is because China is exceptionally good at optimizing performance within any well-understood heuristic framework. What that means is that everywhere the rules of the game are known, China is going to beat the US. They're better, more numerous, more well-organized, and they have a better (collective) attitude.
The simple form of this is "China is good at copying" except that's not really it. It's more like "if China knows the rules at all, it will beat you at the game."
Not to be lightly racist about it, but think of the Asians you know in real life: any time the standards of performance and the parameters of the game are well-understood, they always top the leaderboard.
And because the world is settled, and most of the geopolitical and techno-industrial game is known, China is just going to win. (They have already won at a lot of it; they're just hoping the US doesn't notice)
Don't go thinking "Oh, the US simply needs to pull a heroic turnaround and beat those primitive commies." Those primitive commies are the ones pulling the heroic turnaround. And besides, the US will need to be heroic to pull off even this plan I'm about to tell you.
What China is less good at is the utterly wild innovation that characterizes US culture. The Chinese themselves acknowledge this. They are innovative, but they are not as innovative as the US, which at its best is wildly, incredibly innovative and fearless in the extreme.
China is brave and courageous, the US is wild and crazy. That is the difference.
The problem is that innovations at the margins of current technology are not enough to yield a civilization-scale advantage to the US in this or the next generation. You think you're just going to develop AGI/ASI first? China will have it the next week and they'll have ten of them. It's not going to be in AI.
There is one area where the US still maintains a decisive lead, one that cannot be immediately replicated (relatively-speaking), and it contains the inherent leverage needed to leapfrog the US economy and civilization ahead of all others.
It is reusable self-landing rocketry.
Yes, China is also currently working on this, but it will be at least a couple years before they can do what the US (through SpaceX) can do. Right now, SpaceX on its own is the world leader in both annual launch mass and number of launches.
In Q1 2024 (most recent data I could find; maybe @elonmusk has better numbers), SpaceX itself lofted 429,125 kg. China's CASC was a distant second, at 29,426 kg, less than 7% of SpaceX. SpaceX launches over an order of magnitude more mass than China does. This is the only major area where the US maintains a decisive order-of-magnitude advantage over China.
Not only that, but remember what I said earlier about KPI plus key drivers? The key drivers are what affect the likely rate of change. It doesn't matter if your current state is 2x of your competitor if that competitor is rising 10x as fast. SpaceX is ahead and has the higher rate of growth, and will accelerate once Starship is available.
In other areas, China has the higher rate of growth and is ahead.
Reusable space launch is the key industrial area that the United States has a decisive current advantage in and must now push forward in all ways possible.
Why?
Because unleashing low-cost space launch has allowed the US and US-based companies the exclusive opportunity to achieve asteroid mining.
This is the only outcome within current reach that has a chance of yielding enough payback for the US to leapfrog China.
Successfully lofting sufficient mining equipment, taking it to promising asteroids, mining it for ores and returning it will yield literally trillions of dollars.
And because true power and value are actually about having things, and not money, it's not the trillion of dollars, but the fact that US-based companies will bring back untold amounts of precious and commodity minerals. Even certain single asteroids contain more of these minerals than the entire Earth's reserves of those materials.
(If your reaction is "that will crash the market," your understanding of economics is shallow and you should also just close the window)
US industry will be equipped with and have control over vast amounts of valuable basic materials, at a scale never seen on Earth. Forget Greenland, the current US rocketry advantage gives it a leg up on this race to the most economically valuable real estate in the solar system.
My estimated ultra-aggressive timetable for this puts it at 7 years:
2 years: key technology development and testing
2 years: first mining mission
3 years: scaling and commercialization
(ref: xitter.cf/i/grok/share/s2SkouauW…)
This means that putting the country's efforts on this path has to happen in this Presidential administration.
It's all well and good to have grand plans, but setting them within the scope of that's already understood ("reshoring manufacturing jobs") is not going to work when played against an opponent who is more well-resourced, well-practiced, and operates smartly. If you're going to execute a grand plan to make America Great - truly GREAT - again, you need to pick the right goals for your grand plans.
The US will only overcome China by changing the rules of the game, by opening the arena and finding a new place to compete, where the variables are radically different.
The US is not smarter than China. The US is not more hard-working than China. The US is crazier than China.
A week or so ago, I wrote a half-serious comment about one of the underrated cultural drivers of China's unusual success. Well, America has one too, and it is an asymmetric talent. America thrives best in total fucking chaos, in the wild untamed expanse of the unknown.
I don't know why this is, maybe it's the generations of selection for immigrants (from everywhere, including China itself!) who are willing to cross an ocean and forge a new life with nothing. Americans are just the best in the world at jumping into a totally crazy situation, figuring out how to adapt, and then figuring out how to scale up organized structures so that everyone can thrive in it.
Yes, chaos is painful, and there are many reasons why calm suburbs are nicer but do not forget that this is America's primary and most enduring strength.
The US is not going to beat China in a kinetic war (maybe it will, but do you want to take the gamble?). The US is not going to beat China in an economic war. All of those things are a known terrestrial game, and China is very very good at playing any known game.
But the US has a decisive cultural and technological advantage when competing on a truly new frontier of unknowns. And there just so happens to be one that SpaceX has opened the door to, and handed to the US a decisive early lead - and that is space launch.
I am confident enough in this prediction that I am willing to have people call me on it in a decade. Go ahead, pull out your little reminder apps.
Either the US makes this happen now and brings home a trillion dollars of minerals in the sky and locks down the most promising asteroids, or...
.... it doesn't. It dithers and hems and haws and fights its little culture war, and in 2-3 years China masters reusable rockets and China goes and mines all the asteroids and the future lightcone of mankind is mostly Chinese because believe me, China will construct those space habitats and they will make very good use of all those minerals.
(I don't, like the small-minded hawks, think that China will dominate and oppress the US. I just think China will leave the US to rot, and the US will go the way of the Europeans today)
===
Other details:
Moonbase? Yes, moonbase for refueling and maybe He-3, but it's just something we do on the way. Do not just stop at the Moonbase.
Mars settlement? I love the romance of settling Mars, but the honest story is that it's harder and the economic payback is not as clear. There's not as much on Mars that you can affordably ship back - it's not like New World colonization: even gold (if we found it on Mars) would not be worthwhile to ship back.
The asteroids are different: there's tons of minerals, and they're high up in the gravity well so you can just drop them back down.
Also, I think the US would prefer to be the nation controlling the "dropping back down towards Earth" of very heavy things from up in the gravity well, you know?
===
Disclosure: I am long SpaceX stock
David Li retweeted
It’s been a tough few weeks. My 10yo daughter was diagnosed with a very rare, aggressive cancer called interdigitating dendritic cell sarcoma (IDCS). I’m reaching out to identify clinicians/patients who have encountered pediatric IDCS, indeterminate dendritic cell histiocytosis or other (non-LCH) histiocytic sarcomas cases.
I'm trying to understand non-surgical chemo and targeted therapy options, new pathology markers to better diagnose subtypes/treatments, and any data on progression in pediatric patients. Please feel free to share – I’m trying to cast a wide net due to the rarity of this condition and how little is known.
People can contact me directly at my first name (as written in my profile) at octant.bio.
David Li retweeted
I have different view.
1. Innovation for patients can/should come from anywhere in world.
2. Competition good, everyone works harder, make better or lower cost Rx.
3. US should respond by rewarding innovation + lower time/costs to bring Rx to market. wsj.com/health/pharma/the-dr…
David Li retweeted
ICYMI @davidycli post on X re: 🇨🇳vs 🇺🇸 now available on Timmerman Report. Recommended.
China Biotech is Here to Stay, and How the US Can Continue to Lead. By @davidycli timmermanreport.com/2025/01/…
David Li retweeted
China Biotech's Rise Has Been Remarkable. Here's How the US Can Continue to Lead. By @davidycli timmermanreport.com/2025/01/…
David Li retweeted
One of the remarkable things for me about NeurIPS this year was how quickly the entire AI for Biology community has gone all-in on biological foundation models. Virtual cell models will enable us to predict how cell states will change in response to chemical perturbations. Protein language models will enable us to identify better enzymes for degrading plastics, and so on. Everyone wants bigger data on more things to throw into bigger models.
These models are going to be awesome, but real biology discoveries look somewhat different. Contrast these dreams of foundation models with the latest table of contents from Science or Nature:
--“A long noncoding eRNA forms R-loops to shape emotional experience–induced behavioral adaptation” — The authors identified a lncRNA in mice that is expressed in response to neuronal activity that modulates the 3D structure of chromatin, thereby activating genes that are involved in neuronal plasticity. The authors further identified that this lncRNA is essential for certain forms of learning.
--“Cancer cells impair monocyte-mediated T cell stimulation to evade immunity” — The authors identified that mouse melanoma cells secrete a lipid metabolite that prevents monocytes from activating CD8+ T cells.
--“Postsynaptic competition between calcineurin and PKA regulates mammalian sleep–wake cycles” — By generating mouse knockout lines, the authors identified phosphatases and kinases that are critical for regulating the sleep-wake cycle, and showed that they act through regulation of proteins at excitatory postsynaptic sites.
I struggle to imagine how any of these discoveries could fall out of a multimodal biology foundation model. This is not intended to be a straw man argument. Surely, a foundation model could potentially identify the lncRNA from the first paper, but I am not sure how such a foundation model would associate it with chromatin remodeling. A multimodal foundation model with enough data could also potentially identify metabolic changes associated with melanoma cells subjected to certain kinds of treatments, but I don’t see how that foundation model could identify the effect of those metabolites in preventing CD8+ T cell activation. Indeed, I do not think that any of the foundation models that are being developed today would be capable of generating rich new biological insights of the kind described in these papers. And yet, these are the kinds of insights that new therapies are made from.
The issue, I think, is that machine learning models work extremely well on structured data, and so all the foundation models that are being built are highly structured. Take a protein sequence as input and produce a protein sequence as output. Take a cell state and a chemical perturbation as input and produce a new cell state as output. Biology, however, is poorly structured. The lncRNA insight is case in point: what structured representation can we use for the action of the lncRNA in modulating chromatin architecture? Protein models cannot represent it; DNA models cannot represent it; virtual cell models cannot represent it. Perhaps a model that incorporates RNA expression and 3D genome state could represent it, but then how would that model represent the lipid modulation of the monocytes? I worry that every discovery may need its own representation space. Indeed, the nature of biology is such that there likely is no representation, short of an atomic-resolution real-space model of the entire organism, that is sufficient to represent the diversity of biological phenomena that are relevant for disease.
Except, of course, for natural language, which is evolved to represent all concepts that humans are capable of contemplating. Indeed, I think natural language has an essential role to play in representing biology, and is ultimately unavoidable, insofar as it is the only medium we know of that is sufficiently structured for machine learning and sufficiently flexible to represent the full diversity of biological concepts. At FutureHouse, we work on language agents, which is one way of combining language and biology, but this is not the only way. Models that combine natural language with protein, DNA, transcriptomics, and so on will also be extremely productive, provided the addition of the structured datatypes does not restrict their ability to represent unstructured concepts. However we do it, I think this essential role of natural language in representing biology is currently largely underappreciated.
The history of biology is built on tools that we have found in nature to study biological phenomena. As all biologists know, trying to engineer things from scratch (almost) never works; what works is finding things in nature and repurposing them. It will be aesthetically pleasing if it turns out that our engineered representations are yet again insufficient for studying biology, and that natural language is simply another such tool that we have found in nature that must be applied instead.
David Li retweeted
General Proximity (@proxbio) is coming out of stealth.
General Proximity is developing a new class of medicines that we believe has the potential to decrease global disease burden, transform human health, and extend lifespan more than any other current therapeutic modality. 🧵
David Li retweeted
Competitive landscape for targeted therapies in oncology (TO)
425 assets across 19 pharma companies, segmented by tumor type, phase of development, modality and ownership
If you'd like a PDF copy or if I missed any key assets, let me know below!
Some takeaways & stats:
David Li retweeted
This is the way.
Hype aside: make trials 2-3x more successful, we’ll still fail a lot (this is drug discovery) but we’ll save millions more lives. The current system sucks and even if we end up the same success rate, at least we took a shot to make it better.
"If we take the success rates we observed for AI-discovered molecules in Phase I and II and assume that these hold in the future, and if we combine these with historic Phase III success rates, a striking picture emerges:
the probability of a molecule succeeding across all clinical phases end-to-end would increase from 5–10% to ∼9–18%."
^^^ this is just the very beginning
sciencedirect.com/science/ar…
David Li retweeted
Best thing I've read today.
Market Risk is under appreciated.
Please read 👇🏽
One of the prevalent fallacies about investing in deep tech / biotech is that it’s all about technical risk rather than market risk:
Who wouldn't want better cancer therapies?
If you build it, they will come etc etc
I'm not satisfied with that for 2 major reasons.
1) Sometimes identifying the exact market risk is not exactly straightforward.
The specifics of the product use case matters.
A drug that doesn't clear the current standard of care in a clinically valuable way will not get adopted. A drug that is marginally effective but is multiples more expensive than alternatives will not get adopted. A drug that is quite effective but is extremely difficult to tolerate will likely not get adopted.
In this sense, the importance of identifying what is the market risk is real.
2) Even if you have identified the market risks AND have de-risked the tech risks, that doesn't necessarily make for useful / valuable products
There’s a long distance from a) tech or platform risk being minimized to b) having actually useful products.
Lots of q's to still answer e.g, can the team identify where to point the platform technology in order to generate clinically valuable products? Were the tech risks addressed the right ones to focus on in giving the co a sustainable advantage in building clinically useful products?
IMO it’s not an either or of market vs technical risk.
The specific market spec should drive the technical risk - or which technical problems need to be surmounted to make clinically useful products. Investors (and management) should underwrite accordingly.
Let's keep building 🙏
David Li retweeted
INTRODUCING RAVELIN
I’m excited to finally talk about what I’ve been up to.
Ravelin (@ravelincap) is an early-stage venture firm backing exceptional founders building Critical Software.
Here is who we are, what we do, and what we believe in:
ravelincap.com/introducing-r…
David Li retweeted
Continued incredible progress at @MelioraTx, huge congrats to CEO @davidycli and team!
finance.yahoo.com/news/melio…
David Li retweeted
1/ I can’t believe it’s been a whole year since we launched Powerset, where top technical founders get capital & coaching for angel investing. Today, we’re introducing the nine founders we selected for Powerset Cohort 1… and are opening applications for Cohort 2!
Read more >>
David Li retweeted
How to Do Great Work: paulgraham.com/greatwork.htm…
David Li retweeted
Our portco company is hiring for a Head of Cancer Biology and a Medicinal Chemist!
More info from Founder David Li below 👇👇👇
#hiring #biotech #jobsearch
@MelioraTx is building a computationally powered mechanism-of-action (MOA) deconvolution engine.
We have so covered so much ground in the last several months that we now hiring a Head of Cancer Biology to help spearhead both our pipeline and platform!
David Li retweeted
The next decade will be one of the most extraordinary times for scientific, technological progress
Loved bringing together founders, experts, funders accelerating the speed of research and closing talent gaps w @SchmidtFutures, @Lux_Capital
Next builder dinner: LA. DM for info
David Li retweeted
A founder CEO is an SEO, ie. Sales Executive Officer. 👇 💯
Raw intelligence is important, but overrated, especially when compared to emotional intelligence. Everything you’re doing as CEO is sales related (selling candidates, investors, partners, etc.) — @davidycli