This week has been absolute madness for health tech. If you are building in this space, stop what you are doing and pay attention. The landscape just shifted under your feet.

We have seen the big AI giants finally show their hand, massive M&A activity returning, and deep tech capital unlocking. I have gone through all the announcements from the J.P. Morgan Health Conference (JPM) and the big tech releases so you don’t have to.

See below to continue…

Continue…

The biggest news of the week - and arguably the year so far - is OpenAI acquiring Torch, a tiny four-person startup, for roughly $100 million.

Let’s look at why this matters. Torch wasn't building a "better AI doctor." They weren’t trying to diagnose cancer better than a radiologist. They were building something much more boring and much more valuable: context.

They call it "medical memory." Their software connects the dots between your scattered health records - labs from one clinic, wearables from your wrist, and PDF notes from a specialist - into one unified view. OpenAI is plugging this technology directly into their new ChatGPT Health product.

For the last two years, thousands of founders have raised money to build "chatbots for health." They basically built a nice interface (a wrapper) around GPT-4 and added a system prompt that said, "You are a helpful doctor."

That business model is now dead. OpenAI just killed it. By integrating Torch, ChatGPT now has native access to medical records and context. It doesn't need your wrapper anymore.

Torch won because they solved the data integration problem, not the chat problem. The opportunity now isn't to build the interface; it is to build the "plumbing" that feeds these models. If you can figure out how to cleanly extract data from a legacy hospital system and feed it to an AI, you are sitting on a goldmine.

While OpenAI is winning the consumer, Anthropic is coming for the enterprise.

They just released Claude Cowork. This is a massive shift. Until now, we have been using "Generative AI"- it creates text or images. Cowork is "Agentic AI." It doesn't just chat; it does work. It can access your file system, open folders, plan tasks, and execute them without you holding its hand.

At the same time, they launched Claude for Healthcare. This isn't for you and me; it is for the back office. It connects directly to industry-standard systems like the CMS coverage database and ICD-10 coding systems.

Healthcare administration accounts for roughly 20% of US GDP. It is a bloated, inefficient mess of fax machines, PDF forms, and manual data entry. Anthropic is positioning Claude not as a doctor, but as the ultimate medical administrator. It can check insurance codes, draft prior authorization appeals, and summarize patient histories for the doctor before they walk into the room.

There is a big debate right now about whether AI will replace doctors. I don't think so. I believe we will see the "Jevons Paradox": as technology increases the efficiency with which a resource is used, the total consumption of that resource increases rather than decreases. By automating the admin work, we won't need fewer doctors. We will just consume more healthcare because the friction to access it will drop.

If you are wondering where the exit opportunities are, look at Big Pharma. They are currently staring down the barrel of a "Patent Cliff".

Over the next few years (2026–2028), patents are expiring for some of the world's best-selling drugs. We are talking about blockbusters like Keytruda, Opdivo, and Eliquis facing generic competition soon. This puts an estimated $200–$300 billion in annual revenue at risk.

Pharma companies are panicking. They need to fill that revenue hole, and they have cash to burn.

This is your buyer. You don't need to build a whole pharmaceutical company. You just need to build an asset - a new molecule, a new delivery mechanism, or a better discovery platform - that a panicked Pharma giant can buy to save their stock price.

For a long time, AI in health was just hype. "It will revolutionize everything!" but nobody could prove it. This week, we got the proof.

A major study came out analyzing the Brainomix 360 Stroke platform. This is an AI tool that reviews CT scans to spot strokes. The results were incredible:

Why 64 minutes matters: In stroke care, there is a saying: "Time is Brain." Roughly two million brain cells die every minute a stroke goes untreated. Saving an hour isn't just "efficiency." It is the difference between a patient walking out of the hospital or leaving in a wheelchair.

Stop selling "AI efficiency" or "cost savings." Start selling clinical outcomes. If you can prove your tech saves brain cells, hearts, or lives, the sales conversation changes completely.

Finally, it’s not just about software. "Hard tech" is having a moment.

Quantinuum (a quantum computing company) has filed confidentially for an IPO. They are eyeing a valuation north of $20 billion. This highlights a massive gap between the US and Europe. We have brilliant quantum scientists in the UK and Europe, but the capital markets here just can't support these valuations. US companies like IonQ are trading at billions, while European equivalents struggle to raise at a fraction of that.

But the signal is clear: Investors are willing to bet on deep, scientific risk again. If you are building something hard - quantum, robotics, fusion - the window is open.

If I were starting a company today based on this news, here is where I would look:

Anthropic is going after the big stuff, but there are thousands of niches they will miss. Pick a horrible, specific administrative task. Maybe it's "appealing denied dental claims" or "scheduling MRI appointments." Build an agent that does only that, but does it perfectly. Sell it to the clinics that are drowning in paperwork.

Be the Torch of your niche. The world is full of disconnected data silos.

The stroke AI example proves that you don't need to replace the specialist. You need to help the generalist. Build tools for nurses, paramedics, and GPs. Give them the ability to spot rare diseases or complex conditions instantly. You aren't replacing their judgment; you are giving them a safety net.

Before you get too excited, remember the challenges:

The opportunity is massive. The money is flowing again. But the "easy mode" of 2023 is over. Now we have to build the real stuff.

✅ If this was useful, pass it on. You know a founder still building for the healthcare world of 2023. Send this to them. Show them where the money is actually flowing.

(👉 Vote now — we’ll share the results in next week’s issue. All votes are anonymous.)

(Last week’s poll results)

👉 If you knew for a fact that you only have 3 years left to build before AI takes over, would you still be working on your current business?