The Monday Recap
OpenAI's $11B so far, Fal at $16B, Clutch for $500M, and a rubric for bias in clinical AI
Last week in AI and life sciences: OpenAI reportedly raised part of a new round, Fal doubled its valuation, Palo Alto Networks is in talks to buy an agent security startup, and two health stories asked what clinical AI should learn from and how it should be built.
By Matteo Favilli, 2026-11-09
OpenAI has reportedly raised $11 billion so far in a new funding round that values it at $1.25 trillion. Bloomberg reported it on Friday, and TechCrunch relays that the round has not closed, is expected to close in the coming weeks and could total about $20 billion. In late September Bloomberg had reported a target of $30 billion. Bloomberg also reported that potential buyers of shares in Anthropic's IPO are eyeing a lower valuation than targeted, possibly as low as $1.1 trillion against the $2 trillion investors were said to expect roughly a month ago.1
Fal, a San Francisco startup that hosts and runs AI video, image and audio models, raised $1.3 billion in a round led by Bessemer Venture Partners, with Nvidia, Sound Ventures and T. Rowe Price taking part. The deal values Fal at $16 billion, double its $8 billion valuation in April, according to Bloomberg. Fal announced its Series E and F rounds together, with $1.65 billion raised across both. Bloomberg's report noted that Fal passed $800 million in annualized revenue earlier this year. TechCrunch adds that inference companies such as Baseten, Fireworks AI and Modal are also growing fast, and that their gross margins stay thin because compute is expensive to buy or lease.2
Palo Alto Networks is in talks to acquire Clutch Security, a startup that secures AI agents' access to company systems, for $500 million in cash plus milestone-based payouts, according to a source cited by TechCrunch. Clutch was last valued at $85 million when it raised a $20 million Series A in January 2025, and it is approaching $10 million in annual recurring revenue. TechCrunch puts the price at about 50 times that revenue. If the deal goes through it would be Palo Alto's seventh acquisition this year, per PitchBook. TechCrunch also notes that Cyera acquired Oasis Security, which also manages AI agents' access to systems, for $1 billion.3
In PLOS Digital Health, a team led by Alon Gorenshtein, with Girish Nadkarni among the authors, published a conceptual paper on bias in clinical AI. The authors argue that a model recommending something different when a patient's race or gender identity changes may be showing a justified difference or an unjustified one, and that the origin decides which. Their example: Black race raising the prior for sickle cell disease in a patient with acute chest syndrome reflects how a heritable trait is distributed, while a higher pain-treatment threshold reflects false beliefs that clinicians have been documented to hold. A model trained only to predict records that a feature carries information, never why. They propose a six-criterion rubric, four ways to apply it and an evaluation that reports preserved clinical sensitivity alongside reduced bias. The paper adds no experiments and says its claims are propositions.4
Mount Sinai's Icahn School of Medicine received a 5-year, $4 million NIH grant to build AI models that predict which patients are at highest risk of acute kidney injury after surgery. The models will use data from more than 320,000 patients at 15 hospitals, trained with federated learning, which lets the hospitals train a model together without moving individual patient information between them. MedCity News reports that about 12% of noncardiac surgery patients are expected to develop the condition, and the team plans to test which preventive steps work best for different patient groups. The principal investigator, Dr. Ira Hofer, says clinical trials of the models will start once they work.5
Matteo's take
Three of the five stories are about money, and all three come through reports. OpenAI's $11 billion is reported by Bloomberg for a round that has not closed, and the $20 billion total is a source's figure. The Anthropic valuation range is also what potential buyers are said to be eyeing. The same week, Fal doubled its valuation on $800 million of annualized revenue, and Clutch is priced at about 50 times its revenue in a deal that has not been signed. Large numbers are being attached to AI infrastructure and to AI security, and the sources describe both as busy markets.
The Fal story carries a detail that sits next to those numbers: inference providers grow quickly and keep thin gross margins because of the cost of compute. Revenue growth and margin are different measures, and the report gives the first as a figure and the second only in words.
The two health stories share a question about where a model's output comes from. The Gorenshtein paper says an audit can show that a recommendation changes with a patient's race or gender identity but cannot show whether the change is clinical or inherited from unequal care, so it asks for a named rubric and a record of each decision. The Mount Sinai project starts from a different place: a defined outcome, kidney injury, and a large pool of data from 15 hospitals kept where it is. The first is a proposal with no experiments and the second is a funded plan that has yet to produce a model, so neither tells us yet how well it works.
A clinical risk model like the one Mount Sinai plans will carry the same question the paper raises, because it will learn from records that reflect how patients were treated. The paper's test would apply to it directly: would the association persist if unequal care had been removed?
The concrete thing to watch is the Mount Sinai team's own next step, the comparison of preventive measures across patient groups, since that is where the model's predictions meet differences between groups.
Sources
Drafted with AI from the sources below.
- OpenAI reportedly raises an additional $11 billion at a $1.25 trillion valuation, TechCrunch
- AI inference startup Fal doubles its valuation to $16B as demand surges, TechCrunch
- Palo Alto Networks in talks to buy AI agent security startup Clutch for $500M, TechCrunch
- Context is all you need: A causal framework for governing bias in clinical AI, PLOS Digital Health
- Researchers Receive $4M NIH Grant to Build AI Model To Prevent Kidney Injury, MedCity News
All of AI Bits