Eli Lilly's orforglipron, a once-daily oral GLP-1, is expected to receive an FDA decision by March. Retatrutide, a triple-agonist injectable, follows in 2028. Clarivate projects combined sales of $46 billion by 2031. Morgan Stanley puts the broader cardiometabolic market at $150 billion by 2035.

Those numbers get attention. What they represent matters more.

The current GLP-1 landscape was built around injectables. Clinics structured workflows around injection training. Telehealth platforms differentiated on speed to syringe. Compounding pharmacies scaled to meet shortage-driven demand for semaglutide and tirzepatide.

An oral option changes the infrastructure calculus.

No injection training. No cold chain concerns. No patient hesitation about needles. The operational advantages that some practices built around injectable delivery become less defensible when a pill exists.

This doesn't mean injectables disappear. It means the moat shrinks.

Practices that competed on access, getting patients injectables faster than the branded supply chain allowed, face a different question when supply normalizes and oral alternatives arrive. Access was a position. It was never a permanent one.

The compounding question sharpens here.

503A pharmacies filled a gap created by shortage. That gap funded expansion, capability development, and market presence. But gaps close. Branded manufacturing scales. Oral options sidestep the delivery complexity that made compounding attractive in the first place.

The practices and pharmacies that treated shortage-driven demand as a business model rather than a window are exposed. The ones that used the window to build deeper clinical relationships, broader service lines, and operational credibility have something to stand on when the window closes.

The $150 billion projection isn't a promise. It's a pressure test.

That kind of volume doesn't flow through existing infrastructure without friction. Prescribing capacity. Insurance negotiation. Prior authorization bottlenecks. Distribution systems that weren't designed for this scale.

The question isn't whether demand exists. The question is which systems can absorb it without breaking.

Telehealth platforms are already repositioning. The ones that built around injectable GLP-1s as a wedge product now need to decide whether they're GLP-1 companies or primary care companies that happened to start with weight loss. The answer determines what they build next.

Clinics face a version of the same question. GLP-1 demand brought patients through the door. What keeps them?

The practices that treated GLP-1s as a funnel into broader metabolic care have a path forward. The ones that treated them as the product face commoditization pressure as access widens and differentiation narrows.

None of this happens overnight. Orforglipron still needs approval. Retatrutide is two years out. Branded supply chains take time to scale. Insurance coverage remains uneven.

But the direction is visible.

The GLP-1 market is shifting from scarcity to abundance. From injectable-first to oral-possible. From access as advantage to access as baseline.

The practices that survive the shift won't be the ones who rode the wave longest. They'll be the ones who used the wave to build something that doesn't depend on it.

That's not a prediction. It's already happening.

I’ve been following a project where PharosAI is partnering with 10x Genomics to use the Xenium spatial platform to build one of the most comprehensive multimodal cancer datasets ever assembled. This effort is backed by nearly £19 million from the UK government, with additional support from charities and industry partners, which signals how serious this initiative is.

The goal is straightforward but ambitious.

Turn decades of archived NHS cancer tissue into high-resolution, AI-ready datasets that can actually be used to improve diagnosis, guide precision therapies, and accelerate drug discovery.

What makes this different is the depth of data being combined. Genomics, transcriptomics, imaging, and spatial biology are all being linked together and paired with custom AI models.

That combination allows researchers to surface patterns in cancer biology that were previously invisible.

The collaboration itself is equally important. This is a joint venture across leading academic institutions and NHS trusts, bringing together real clinical expertise, academic leadership, and advanced technology under one roof.

It is not theory or bench science in isolation, it is deeply connected to patient samples and real-world care.

The program is launching with a major focus on breast cancer, then expanding into lung and pancreatic cancers, with plans to include thousands of clinical tissue samples.

To pull that off at scale, they needed a spatial platform that could deliver consistent, reproducible results across massive archival datasets.

Xenium’s throughput and flexibility, including the ability to build cancer-specific gene panels, made that possible.

What I find most compelling is the commitment to access. These datasets will not be locked away. They will be securely shared with researchers and innovators globally, creating a foundation others can build on.

Running through 2027, this initiative puts the UK at the center of AI-driven cancer research and creates a real ecosystem for discovery in precision medicine. This is how earlier detection, smarter therapies, and faster breakthroughs actually happen.

I've spent the past decade building and scaling operations across clinics, telehealth platforms, and compounding pharmacies. I've seen practices succeed at 20 patients per month and break at 200. I've watched regulatory shifts reorganize entire markets overnight. I've built the infrastructure that determines whether modern therapies scale or stall.

This newsletter is about understanding how the system actually functions, where friction hides, and why some approaches compound over time while others don't.

If you're building, running, or participating in modern health in any real way, this layer matters more than most people realize.