Daniel Ek targets US medical system with $499 AI scans

American healthcare burns 18 percent of national GDP on treating illness after it strikes.

Speaking on All-In, Spotify co-founder Daniel Ek pointed to a structural misalignment in how medicine is financed. Because the average American worker switches jobs every two to three years, insurance carriers face no incentive to fund early intervention. A preventative treatment with a decade-long payback window offers zero financial return to an insurer whose policyholder will transfer to a competitor long before disease manifests.

Ek founded Neko Health in 2018 to bypass insurance gatekeepers with direct-to-consumer data collection. For $499, the clinic conducts a one-hour assessment combining 53 blood biomarkers, cardiovascular testing, grip strength evaluation, and 6,000 high-resolution camera images indexing every millimeter of a patient's skin.

The visual dataset solves a basic human limitation. The average adult carries roughly 950 moles across their body, making manual tracking across annual doctor visits nearly impossible. Neko uses artificial intelligence models to track microscopic changes and flag anomalous growth patterns year-over-year, handing those visual flags to clinicians to detect early melanomas and unseen cardiovascular risks.

Early results across Europe show the model finding hidden illness at scale. In 100,000 patient scans conducted in Sweden and the United Kingdom, Neko identified severe, previously undiagnosed conditions in roughly one percent of participants. Ek noted that members starting with the poorest baseline health showed the most dramatic clinical improvements after receiving their baseline data.

The broader strategy mirrors Ek's playbook at Spotify: build a fine-tuned technical pipeline that lowers unit costs until market behavior shifts. Beyond medical diagnostics, Ek argued on All-In that regulators targeting artificial intelligence are focusing on the wrong target by trying to restrict open-source code.

Drawing parallels to cold-war supercomputer controls, Ek argued that meaningful AI regulation belongs at the datacenter level rather than inside consumer software. A developer running open weights on a desktop cannot trigger catastrophic risks, whereas 100,000-GPU compute clusters demand real infrastructure governance. Spotify itself pairs fine-tuned open models with frontier systems to run personal recommendations without blowing out processing budgets.

The fix for broken systems comes down to hardware economics.