Insights / Technology
Field notes on Foundation Models — the general-purpose models reshaping life sciences, from protein design to clinical reasoning.
20 posts in Foundation Models
6 Aug 2026 · 3 min read
Icon just wired Anthropic's Claude into its clinical trial platform across four production capabilities, and I think this validates my April argument that CROs combining frontier AI with operational data and global execution get more valuable — not less. The commercial question is still open, but the switching-cost dynamic I underweighted in April may be the most consequential part of this deal.
21 Jul 2026 · 2 min read
Bristol Myers Squibb just claimed the most powerful AI supercomputer in life sciences—the third pharma company to do so in nine months—and I think the real signal isn't the superlative, it's the demand curve behind it.
23 Jun 2026 · 2 min read
Roughly 11,500 Novo Nordisk clinical-trial patient records are allegedly on a leak site, next to the recipe for Ozempic — and the breadth of what was taken is the real story. I argue that the same frontier AI pharma has spent two years asking what it can build is now sharpening the tools that make a terabyte of stolen files instantly usable.
16 Jun 2026 · 1 min read
A Nature Medicine study found OpenEvidence, a clinical AI now valued at $12 billion, scored no better than free Google Search on blinded real-world physician queries, while general frontier models led. The sharper lesson for anyone buying clinical AI: which benchmark you trust decides the winner you see.
9 Jun 2026 · 2 min read
Anthropic benchmarked Claude Opus 4.7 against ChemDraw and MestReNova on NMR prediction and structure elucidation, and the general model matched the specialist tools. The read for techbio: encoded domain knowledge is becoming a default capability of general models.
14 May 2026 · 4 min read
I spoke at the Global Health Exhibition Capital Xchange webinar opened by the Saudi Ministry of Health, and the headline story in healthcare venture this year is concentration — less capital, larger checks, AI dominating in techbio and clinical workflows, and a Saudi market that is shifting from consumer to producer economy.
12 May 2026 · 4 min read
Isomorphic Labs is in advanced talks to raise more than $2B, and I'm of two minds about it. The capital is welcome, but the deeper question is whether tech boards will stay patient through a decade of clinical iteration — and whether anyone is funding the downstream bottleneck that more AI-generated candidates will only tighten.
14 Apr 2026 · 3 min read
The hyperscalers are moving on drug discovery from every direction — Anthropic acquired Coefficient Bio, and AWS just launched Amazon Bio Discovery with 40+ biological foundation models and integrated CRO partners. I still don't think the commercial model of pharma is under major threat, but the R&D part is clearly heating up massively.
7 Apr 2026 · 3 min read
Boehringer Ingelheim's new AI and machine learning center in London signals something more specific than another pharma AI announcement — pharma is starting to build permanent capability hubs, not just buy software deals. My bet is that by 2028, the companies with serious in-house computational capability will look structurally different from the ones that outsourced both talent and hardware.
19 Feb 2026 · 1 min read
The AI-in-healthcare debate keeps swinging between “AI is going to take over everything” and “AI is useless in a medical setting.” Neither is useful. What the field actually needs are external neutral benchmarks on specific clinical tasks. Congrats to Rory Davidson and team for bringing one to market.
20 Nov 2025 · 1 min read
Jeff Bezos just bet $106M on Profluent, a company using frontier AI models to design proteins that don't exist in nature — and I break down exactly what makes these models 'frontier' and why it matters for programmable biology.
28 Aug 2025 · 1 min read
Consolidation in healthcare AI is inevitable, but the $81M acquisition price of Paige AI by Tempus AI — after raising over $260M and once valued at $600M — is a sobering reminder that building a great product and figuring out how to monetize it are two very different challenges.
15 Jul 2025 · 1 min read
Participating in the SNOMED International board meeting in Manchester reinforces my conviction that SNOMED CT is absolutely critical for healthcare AI. As a board member chairing the Technology and Innovation Committee, I see firsthand how this global standard transforms fragmented health data into a reliable, structured foundation for AI innovation.
8 Jul 2025 · 1 min read
A recent Microsoft study suggests AI outperforms doctors at diagnosis, but only under artificially constrained conditions — the real story is about AI augmenting physicians, not replacing them. The greatest barrier to better patient outcomes isn't physician skill or technology potential; it's the clinical inertia and outdated infrastructure keeping them apart.
3 Apr 2025 · 1 min read
Google DeepMind's spin-off Isomorphic Labs has just raised $600M from Thrive Capital and GV, the latest mega-round in the blazing AI-enabled drug discovery space. I'm watching closely to see whether bigger funding truly translates to tangible clinical results in this rapidly emerging field.
28 Jan 2025 · 2 min read
Everybody is talking about DeepSeek and how it just erased a sizeable chunk of NVIDIA's market cap — so I asked DeepSeek R1 itself what this means for AI's future energy consumption. The answer is nuanced, and I'm quite impressed with what it came up with.
14 Nov 2024 · 1 min read
I'm seeing remarkable productivity gains using AI daily, but some industry insiders are signaling we may be approaching the limits of the current 'scale-up' approach to LLMs. Here's my take on the three emerging pathways that could define what comes next.
17 Sept 2024 · 1 min read
OpenAI recently unveiled o1 — nicknamed Strawberry — which some reports suggest scores an IQ of around 120, putting it on par with or above many humans. I explore what makes this model different and why its ability to "think" before responding could be a game changer in expert-level fields.
2 Aug 2024 · 1 min read
Microsoft now formally lists OpenAI as a competitor, despite having invested $13 billion into the company — a move that may be tied to SearchGPT or antitrust optics. The strategic manoeuvres around these massive LLM-race investments are captivating.
25 Jul 2024 · 1 min read
I'm tracking the continuous emergence of new AI foundation models — from Mistral Large 2 to Meta's Llama 3.1 — as a billion-dollar arms race. My prediction: three types of models will emerge, and for healthcare, specialised imaging models like Microsoft's collaboration with Mass General Brigham are the ones to watch.