Noor’s Newsletter — Issue #14
This isn’t a recap of the weeks—it’s an attempt to understand the forces reshaping how we live, govern, and evolve.
As 2025 draws to a close, the AI–healthcare–life sciences intersection appears more polarized and more mature than at any point in the past decade. December’s final weeks revealed a field consolidating around two parallel realities: the institutionalization of AI in healthcare systems and regulation, and a sharp recalibration within AI-native biotech as capital, timelines, and scientific risk are reassessed.
AI Applied to Clinical Outcomes
In England, the NHS rolled out an AI-based forecasting system to anticipate emergency department demand during the winter surge. By modeling admissions the system aims to smooth capacity constraints and reduce waiting times. While modest in technical ambition, the deployment is notable for its scale and for the political framing around AI as a public-sector efficiency lever rather than a clinical decision maker.
In the United States, governance took a more operational turn. CMS confirmed plans to pilot AI-assisted prior authorization for a subset of Medicare outpatient procedures beginning in 2026. This expansion into administrative decision-making raises a different class of ethical and policy questions — less about accuracy, more about power, accountability, and appealability.
Techbio ecosystem
If healthcare delivery showed signs of stabilization, AI-native biotech told a more turbulent story. Perhaps the most emblematic development was Verge Genomics’ decision to discontinue its sole clinical-stage ALS program and pivot back to its AI drug discovery platform. Once positioned as a poster child for AI-first neuroscience, Verge’s retrenchment underscores a hard truth confronting the sector: clinical validation remains the ultimate arbiter of value. The decision, however, is somewhat counterintuitive given the prevailing shift within pharma — and increasingly within techbio — toward recognizing the durability of platforms that have already been de-risked through clinical assets. Only a few years ago, the industry actively debated the optimal business model for techbio companies. That debate now appears largely settled. Irrespective of the sophistication of the underlying technology, value in drug discovery continues to crystallize around assets in the clinic. Platforms may enable discovery, but it is clinical progress that ultimately converts promise into value.
At the opposite end of the spectrum, Chai Discovery closed a $130 million Series B at a $1.3 billion valuation to advance its foundation-model-driven approach to protein design. The scale of the round — and the valuation it supports — indicates that investors remain willing to underwrite substantial research and platform risk. Notably, 2025 emerged as a breakout year for protein design, with multiple companies advancing generative models for molecular engineering and securing outsized rounds of capital, signaling renewed confidence in foundation-model-driven approaches to biology.
These contrasting trajectories — Verge stepping back from the clinic, Chai doubling down on foundational models — illustrate the current bifurcation of AI biotech. Capital is consolidating around fewer, early technological innovations, while tolerance for early clinical disappointment is shrinking. The market appears to be demanding either genuine general intelligence at the molecular level, or clear near-term clinical signal — increasingly unwilling to fund the ambiguous middle.
Taken together, December’s developments close out 2025 with a sense of sober momentum. AI in healthcare is becoming operational, governed, and unavoidable. AI in biotech is becoming more selective, more capital-intensive, and more honest about its limits.
The coming year is likely to mark the long-anticipated inflection point for AI-discovered drugs, as a growing number of programs finally reach meaningful stages of clinical validation. Protein design, still early in its research maturity, will continue to move into the mainstream, with an increasing flow of assets transitioning from discovery into development. As modeling and simulation capabilities advance, the impact will extend well beyond antibodies. Improvements in computational accuracy are poised to reshape small-molecule and RNA-based modalities as well, with data-driven simulations increasingly rivaling — and in some cases surpassing — traditional physics-based approaches.
As the field enters 2026, I remain optimistic about our collective ability to advance science in service of human health and to continue making progress toward a future in which disease is increasingly manageable.
On that note, wishing you a happy and healthy New Year!
-Noor
P.S. I've been learning and writing about the things that fascinate me most, and I've compiled them into a book: The Great Rewrite. The first part is now ready, focusing on, not surprisingly, innovations in healthcare, AI, and biology. I'm opening access to the individual book parts, which can be downloaded here: https://www.noorshaker.ai/publications.html. More chapters coming soon. The final book will be available on Amazon shortly.