Particular focus on bio, fintech, deeptech and broader vertical AI themes.
Joined March 2025
- Tweets1.3K
- Following806
- Followers755
- Likes1.4K
Everyone argues about lab valuations, in particular with the upcoming $ANT IPO. The number that matters is the gap between what a token costs to serve and what it’s sold for.
Right now prices absolutely do NOT reflect cost, and the whole application layer is built on that subsidy.
Two forces collide from here. Open-source, increasingly from China, is commoditising base models, which pushes prices down.
At some point the labs burning billions have to charge what inference actually costs, which pushes prices up.
The resolution likely isn’t “bigger models”.
It’s architecture, with routing across many smaller models, cheaper modalities, world models trained on video and physical data rather than scraped text. So the test for any AI company should be simple.
Does the business still work if your model bill triples?
If the answer depends on one vendor’s goodwill, you don’t have a company, you have a reseller agreement.
The founders who treat model cost as a design constraint today will own the application layer when the subsidy ends.
$QURE A few under-discussed positives in UBS’s October 14 note. The most interesting is how few annual treatments could support substantial revenue, and how quickly it could start funding the rest of the pipeline.
The commercial capacity argument is actually stronger than it would first appear.
UBS uses the existing deep-brain-stimulation network as a reference, and specifically outlines that approximately 280 centres performing just three AMT-130 procedures each year would mean 840 annual treatments.
At its assumed $2.5 million list price, that represents approximately $2.1 bln in US sales before discounts. Three procedures per centre is one every four months.
AMT-130 can remain a highly specialised procedure and still become a substantial franchise. The practical work is qualifying centres, securing reimbursement and converting referrals into treatments. Existing DBS infrastructure provides a starting point - those centres would still need to become AMT-130-ready.
UBS models a relatively early transition to profitability.
Its forecasts include approximately $457 mln of revenue, $90 mln of operating profit and $88m mln of operating cash flow in 2028. Operating cash flow rises to approximately $309 mln in 2029.
Those figures depend on a successful launch and the assumed margins. But the implication deserves attention. AMT-130 could allow uniQure to finance further development from operations, materially reducing its dependence on issuing shares.
That creates a potential second stage to the investment thesis after approval:, i.e. a growing business funding its own pipeline.
AMT-260 gives the broader CNS thesis something concrete to build on.
The epilepsy programme targets different biology in a different brain region, using gene silencing to reduce pathological excitation while preserving tissue.
Stronger subsequent results could demonstrate that uniQure’s brain-delivery and development capabilities are repeatable across diseases. UBS currently only assigns this programme only a 30% probability of success and roughly $3 per share of value.
The biotech vertical is real.
One-time CRISPR edits cutting HAE attacks 87% and LDL by up to 62% (holding a year+) prove durable genetic fixes work in humans.
That de-risks the space for $QURE. Brain delivery is the hard part, and they’ve done it.
bio/acc is arriving for one-and-done gene therapy.
this might be my last normal tweet before biotech goes vertical
i'm not a whistleblower. I don't work at Lilly, Novo or any of the big labs
but there is a massive shift happening in human biology right now.
it's sitting in press releases and almost nobody reads them
the last 6 months alone:
- a once-daily pill roughly doubled median survival in metastatic pancreatic cancer (13.2 vs 6.6 months)
- a one-time gene editing infusion cut hereditary angioedema attacks 87%, and 62% of patients stayed attack-free and off all other prevention for 6 months
- a one-time gene edit for cholesterol dropped LDL up to 62%, and it held for over a year
- a personalized mrna vaccine beat keytruda alone in a 1,137 patient melanoma phase 3
- retatrutide hit ~28% weight loss in a phase 3
- an ai-designed drug entered phase 3
- a pill that regrows hair passed its second phase 3
- an immunotherapy that could let some rectal cancer patients skip surgery, chemo and radiation just got fda approved
- a man lived 271 days off dialysis on a pig kidney
- human cells were caught passing chromosome-sized chunks of DNA to their neighbors, and it changes how the neighbors behave
we are on the verge of an era where the human body is a software problem and the update cycle gets shorter everyday
stick around, it's going to get weird
bio/acc
Rest in peace, Sir Patrick. You will be missed.
There was something extraordinary about the way Sir Patrick Stewart could make even the most powerful characters feel profoundly human.
Whether commanding the bridge of the Enterprise or guiding a generation of young mutants, Sir Patrick Stewart brought a tenderness to his characters that made us feel we knew them. Behind that unmistakable voice was an actor who understood that even the greatest heroes could be uncertain, that strength could exist alongside vulnerability, and that compassion was never a weakness.
Over a career spanning more than six decades, his performances became woven into the lives of generations who grew up watching him. As Captain Jean-Luc Picard, Professor Charles Xavier, and so many others, he left us with characters we loved, lessons we carried, and memories we'll always treasure.
Prescient comment from Thiel
Peter Thiel might be onto something here…
aleximas.substack.com/p/what…
FWIW I think he’s missing important dimensions like desire for human connection, empathy, shared understanding, which will be important for the future of jobs in healthcare, education, etc.
ClearPath retweeted
Not many people seem to know about Metaplanet, Jaan Tallinn's VC firm.
They are the most AGI-pilled firm on the planet, and are about to flip many of the tier 1s.
> Invested in @GoogleDeepMind in 2011
> Led @AnthropicAI series A in 2021
> Doubled down on Anthropic in every round since. IRR hurdle for deploying capital into anything else was {absurdly high number}% for years
> Anchored the premier AGI hedge funds (+Mythos) + in-house public equities strategy
4-person team in Estonia now managing 10s of billions AUM, liquid after $ANT IPO.
Rationalist roots --> zero bureaucracy, cares about human survival & flourishing as north star, doesn't care who else is investing or how high status a founder / idea is
Founders: congrats! you no longer need to play status games to raise a mega round for your civilization-scale moonshot. You can just pitch @metaplanetfund
ClearPath retweeted
Revenue per GPU can increase as token prices decline if per GPU token throughput increases enough.
So both Anthropic revenue *and* margins can go up as token prices go down *if* GPU prices go up less than the product of (token price decline * per GPU throughput increase).
But I have super enjoyed all the commentary to the contrary!
Strengthens the scientific case for using Enroll-HD to understand TFC progression, and importantly the case for TFC being the endpoint.
It could also help the confirmatory trial, as better baseline adjustment can improve precision.
One of the strongest scientific arguments supporting TFC natural-history controls has just been reinforced by independent researchers at Roche, right as $QURE approaches a pivotal regulatory decision.
The argument that Enroll-HD is inherently unsuitable as a TFC comparator is becoming increasingly difficult to defend.
Btw, this is central to the reimbursement case for AMT-130.
ClearPath retweeted
Adam the contrast was the point. Two different bad/fatal diseases yet very different response from the field to advances. We do not have a functional cure for many cancers but HR < .7 in any cancer is described as practice changing in oncology. Neurology seems underwhelmed by the same degree in disease change.
Of course cancer is not one thing and I could have been more precise- CLL is not good but now mostly not fatal, due to many advances from Lilly and our peers HR+ HER - breast cancer has long survival now yet GI cancers are mostly unsolved and survival curves are terrible.
You know what else is terrible? Almost every neuro-degenerative disease including AD and we (including the media) need to discuss it that way. Unfortunately I’ve witnessed end of life journeys up close in both diseases, while different horrors present themselves in each I’d say they are both pretty bad. Advancement 30%+ at a time should change practice and routinely be offered to patients and families. Oncology does this Neurology doesn’t.
$LLY Kisunla is an Alzheimer's drug. It's not a cancer drug. Different diseases, different trajectories, different patient population with different needs, different benefit-risk profile.
@DaveARicks, speaking to @bradloncar, points to the hazard ratio from Kisunla's Phase 3 studies and says, "If this was a cancer drug, it would be a giant drug!"
Ok, but it's not.
1/2 $QURE has a potential second act that deserves more attention.
AMT-130, and in particular BLA acceptance, is clearly the immediate catalyst. But a successful HD programme could establish something valuable beyond one medicine i.e. experience developing, delivering and monitoring durable gene silencing in the human brain.
The investment question is how much of that expertise can be reused.
That is why AMT-260 for epilepsy also matters more than most realize. Stronger results would give investors evidence across a second CNS disease, with a different therapeutic target. The argument for assigning value to the broader platform would become materially stronger.
There is also an interesting scientific building block in linQURE, and that is published preclinical work demonstrated simultaneous silencing of multiple target transcripts from one AAV construct.
AI could help explore that design space, specifically target combinations, therapeutic sequences, unwanted effects. The opportunity is to improve the quality and speed of candidate selection, subject to experimental validation.
2/2 The partnership that would materially strengthen my $QURE thesis is one where big pharma pays uniQure to develop several more CNS medicines.
There is in fact a precedent for meaningful capital flowing into this area.
Roche expanded its CNS collaboration with AI-driven capsid developer Dyno with $50 mln upfront. Dyno also works with NVIDIA on biological sequence design.
Novartis committed $100 mln in upfront consideration to a Voyager collaboration covering Huntington’s and spinal muscular atrophy.
The commercial definitely appetite exists.
For QURE, the details would determine the value: upfront consideration, funded development, retained rights and economics across multiple programmes.
The direction of payment is particularly revealing. QURE paying for a better vector could improve its pipeline. A partner paying for QURE’s capabilities would establish an external valuation of those capabilities.
Any QURE–Dyno or other collaboration clearly remains hypothetical, but still worth exploring.
But AMT-130 regulatory success, along with strong AMT-260 data and a substantive multi-programme partnership could change the market perception meaningfully and also broaden the investor base.
That is the interesting $MRNA-style possibility actually - investors could begin assigning meaningful value to the medicines their platform will produce next.
ClearPath retweeted
Cool!
In partnership with Vinod Khosla (@vkhosla), I’m proud to introduce Reservoir.
Billionaires have owned their cells for years. Starting today, you can too.
Sign up today: withreservoir.com