Big Tech companies in healthcare: losing, reproducibly

OpenAI has recently announced its healthcare ambitions. Why does it feel like we have been here before? Because we`ve been here before. 

For almost 2 decades now, Big Tech has tried to enter healthcare through many different angles. And such initiatives failed organically, with striking consistency.

18 years of data: organic success is notably absent from the dataset

Since the mid-2000s, every major Big Tech player has launched an organic healthcare initiative: Microsoft, Google, IBM, Amazon, Apple.

Here is how it goes: 

• A big announcement.

• A newly formed healthcare division.

• High-profile hires, usually a physician or a scientist.

• A wave of press coverage - arguably the one area where Big Tech consistently outperforms everyone else is press/marketing. 

• And then, a few years later, shut down.

• The median survival across these initiatives is 3.4 years.

• Of the initiatives disclosed, roughly 83 % were shut down, 17% divested. None achieved their original strategic objectives.

Here is a non-exhaustive list:

• Microsoft HealthVault ran for 12 years and was shut down.

• IBM Watson Health survived 6.8 years before being divested to private equity.

• Google Health was launched twice, first lasting 3.5 years, the second just 3.0. Both were closed.

• Haven Healthcare, the joint effort by Amazon, JPMorgan and Berkshire Hathaway, lasted 3.1 years, before being shut down. 

• Amazon Care was shut down after 3.3 years.

Honestly, to find any organic success, I had to travel back to 1994. There are readers of this newsletter who were not yet born in 1994 – that`s how far back it is. That`s when Oracle developed their clinical trial solution, used to this day. However, to build a real business, Oracle acquired many technologies in clinical trials on top. 

Apple Health: there is reach. But where are results?

Apple Health is often brought up as the counterexample, and it is worth addressing with nuance.

On the one hand, Apple Health is everywhere. Hundreds of millions of iPhone users track steps, heart rate, sometimes sleep. It is a massive data collection effort, and the Apple Watch has clearly been a commercial success.

On the other hand, by Apple’s own standards, this is underwhelming. After years of investment and unmatched distribution, the dominant use case remains step counting. Occasionally clinical studies point to the Apple Watch’s FDA-cleared ECG function, but it’s far away from health system integration at scale . Yet there was so much potential. 

Apple succeeded where its core strength lies - hardware. Apple Health, as a broader healthcare platform, never came close to fulfilling its potential. A lack of sustained focus outside the core business might be one of the reasons, but not the only one.

What happened? Entry barriers happened

So how did companies that large fail so reproducibly, and lost to much smaller specialists? 

No, I do not think it is about the focus. You know why? Let`s travel back to 2010s. There was plenty of focus: billions spent organically on healthcare, high-profile hires, press coverage. Something else happened. And I believe it is the good old entry barriers. Healthcare has entry barriers that Big Tech systematically underestimates. 

Why? Because healthcare is not a consumer market. It is not B2C, not even B2B. It is mostly B2B2G (government).

The ultimate payor is not the user, or businesses trying to sell ads to that user, or a business buying cloud services. It is the state, society, and population-level insurers (yet operating under governmental oversight). Yes, there are decision-makers in between, but the end payor is fundamentally different from consumer and B2B technology.

Healthcare, much like defense or energy, is survival-critical. And interestingly, Big Tech has also failed to build organic businesses in defense or other regulated industries (why would Palantir exist at this scale if Big Tech was so great in regulated industries)?

Consumerization exists,  but many B2C niches are too small, unstable and ephemeric

To be fair, parts of healthcare are consumer-facing.

Direct-to-consumer brands like Hims & Hers, show that some niches behave like consumer markets. The rise of GLP-1 therapies has created an entire sub-economy within healthcare that follows consumer rules.

However, let`s take a decades-long view on these B2C niches: they have been mostly unstable. Genetic testing, Fitbit (Apple) even some parts of GLP-1 / incretin receptor agonists are good examples or quickly eroding competitive advantage, which results in unstable business environment. More stable niches have been formed where entry barriers are immense: in medical aesthetics (Galderma) and in over-the-counter medication.

Also, most healthcare budgets still flow into where value is delivered, the physical environments: delivery of care, efficacious medications, life-saving MedTech, IVD, Tools. Software can create very large companies in healthcare, but only when it plays together with the physical infrastructure (remember, our bodies are physical). Also, only software that taps into the B2B2G nature of healthcare scales – because they tap in real business, not imaginary business like many primary care apps. 

This is why the leverage of software in healthcare has been repeatedly overestimated, including by venture capital investors, but equally by buyout or private infrastructure roll-ups. 

If Big Tech lost, who won? Specialists did

The companies that dominate healthcare infrastructure are mostly unknown to consumers. They are:  Epics, Cerners, Genedatas, Informaticas of this world.

Most consumers have never heard of them, just as most people have never heard of Palantir despite its dominance in defense. That is the nature of true B2B2G infrastructure winners.

Healthcare rewards focused specialists who are born in their customers` world , and sign long-term contracts that often transcend geographies.

These are not consumer growth stories. They are infrastructure-like wins.

The Genedata case: where Big Tech missed the real business

If you look at life sciences AI from an academic lens, AlphaFold dominates the conversation. But from a profit pool perspective, the picture looks very different.

• Firstly, AlphaFold came from Google’s acquisition of DeepMind in 2014, reinforcing a broader pattern: healthcare success at Big Tech is almost never organic.

• Secondly, even today, the commercial success of AlphaFold relative to acquisition cost and ongoing investment is debatable – when you have been in investing long enough, it becomes difficult to imagine Alphafold earned its cost of capital.

• Most importantly, Google left enormous profit pools untouched. Yet, they would have been such a classical match to its strategy.

The core API infrastructure used in pharmaceutical drug discovery does not run on any Big Tech systems – and never has. It runs on Genedata, a Swiss specialist headquartered in Basel, built over decades in close proximity to global pharma clients. In 2024, Genedata was acquired from its founder by Danaher, a life sciences tools giant.

This was a missed opportunity for Google to embed itself deeply into pharmaceutical workflows. How do I explain this? Why do Big Tech efforts look so stochastic, missing profit pools? After over a decade in front of Pharma, Biotech, MedTech, Hospital leadership: the strategic insight into where value truly sits in healthcare requires being “born” in the industry. 

Big Tech can be successful via M&A

Every meaningful success came through acquisition:

• Google acquired DeepMind (convolutional neural networks, CNNs) in 2014, and Alphafold became of key life science products.

• Amazon acquired privately-held PillPack in 2018, forming the backbone of Amazon Pharmacy.

• Oracle acquired EHR/EMR leader Cerner in 2022 from the stock exchange. 

In all cases, the healthcare capability already existed. M&A helped Big Tech overcome the entry barriers.

Not all acquisitions succeed. Amazon’s purchase of One Medical shows that even buying expertise does not guarantee success, particularly when the business sits far from the acquirer’s core strengths.

Is this time different for foundational tech, like OpenAI and Mistral?

OpenAI’s and Anthropic’s loud entrance in healthcare has caused uproar in the industry. Some would argue a foundational LLM company has inherently different abilities to expand, innovate and partner than previous Big Tech companies like Amazon. Does that warrant a different buy-or-build strategy? Stay tuned for our next analysis.