There’s about to be a lot of AI capital incineration

I do believe there will be winners in the AI space. There will absolutely be platforms plays here that succeed. OpenAI has amassed billions of dollars in capital and some genuinely impressive technology. I remain bullish on Nvidia. Abridge seems cool. Some of these venture-backed companies in health care will return capital to investors in the next five to ten years.

But things have become borderline unhinged in VC land. I have been talking to a bunch of founders recently who have been telling me that investors will not meet with them if there’s no AI involved. Not even a first meeting! Venture capital firms are increasingly rebranding their thesis as AI-only, and companies are throwing AI into their pitch decks whether they’ve built it or not. It is AI, around the clock, all the time. And everyone needs to have a story about how their solution is powered by it. This dynamic reminds me of the Web3/crypto frenzy, but it’s arguably a lot worse.

Meanwhile, AI is still in its infancy when it comes to impact — especially in health care. Ironically, where I most often see it is baked into virtual clinics that are heavy on services revenue - but have a team of engineers. These companies have an edge because it’s relatively easy for them to incorporate AI into the workflow. Where I struggle to see (most) companies succeed is if they’re selling AI software, particularly if they’re targeting health systems with razor thin margins and enormous workforces.

Another problem is ROI. Some of these early AI applications seem most helpful in improving quality of life for clinicians, and making health workers feel more productive. And those are tremendous steps forward. But I’m not hearing from clinicians that everything has changed when it comes to their practice, or that they have many more to dedicate to their patients. If anything, many of them have only increased the volume of patients in their panel.

It’s also not clear to me that patients are desperately calling out for AI. I truly believe that most of us still want to talk to a human being when staring down the barrel of a massive medical bill that could bankrupt us. Or imagine seeing a scary looking lab value. Do you want a chatbot to tell you that you might have cancer? (If I’m wrong about this, I’m all ears!)

I’m not an AI skeptic. But I’m not ready to throw in my lot with it and stop investing in healthcare services. AI is not the only answer to all of our problems - and we should stop acting like it is.

Here’s where I’m especially bearish — and it’s an argument I’ve made before in this newsletter.

AI companies will struggle if they (and their investors) refuse to acknowledge that to sell into health care, you can’t ignore services. A lot of investors these days are trying to make Tech/SaaS happen, and it’s a perpetual struggle. There aren’t many pristine, super high margin SaaS businesses to be found.

Until there are plans to extensively train workers/teams at payers in how to use the AI technology (at health systems, plans, and so on), there has to be services. These companies will need to have their own teams of humans to react, respond or information. There is a reason why most of the public-traded “tech” companies in health care are heavy on services (see: Health Catalyst, for example). Healthcare startups often have a truly diverse revenue mix whether we like it or not.

“When you have a workforce that turns over very often, and is stressed with the existing ways of doing things without a services front end, you’re not going to get people running towards the use of these AI enabling technologies — you’ll get people running away from them,” Tom Cassels, a friend and colleague, who advises digital health companies on business strategy, shared with me.

I recently attended a conference where health system executives shared anecdote after anecdote about how entrepreneurs with generative AI tools had approached them, but they couldn’t afford to spend yet more money on technology. They don’t have massive budgets, for the most part, and moving their extensive labor forces around (or cutting jobs) isn’t easy for all sorts of reasons, including highly political ones.

What tends to sell most easily into health systems now is mission critical applications - those that help them bolster revenue by filling beds, for instance.

Where this brings me to is that we are overdoing it on AI and missing out on opportunities to invest in solid, services businesses that are truly helping patients. I’ve been seeing some great seed-stage businesses of late that have grown revenue quickly serving patients and selling into entities like Medicaid. These businesses have extremely low valuations and are struggling to raise, relative to their peers in AI that have made zero traction.

This could be a great time for investors to make a truly contrarian bet - and right now, it’s against the grain to bet on human beings. But for the foreseeable future, whether the tech world likes it or not, health care is still rooted in human relationships. We’ll have to be extremely thoughtful about how we bring in AI, including how we invest in it.

I’d love to hear your thoughts on today’s column! As always, drop me a line at christina@secondopinion.media