Startups

Daniel Ek putting Spotify-scale capital and credibility behind a preventative health scanning startup is a bet that consumers will pay out of pocket for diagnostic imaging the NHS and most insurers won't proactively offer. The raise values Neko Health as a serious infrastructure play, not a boutique clinic chain, with ambitions to build scanning centres at a density that makes the model work on throughput rather than premium pricing alone. The tension is regulatory: body scanning at consumer scale raises exactly the kind of overdiagnosis and false-positive concerns that have made the NHS and equivalent European systems cautious about population-level imaging. If Neko can show its scans reduce downstream healthcare costs rather than just generate anxious follow-up appointments, it has a real moat. If it can't, this is an expensive bet on the worried well.

Releasing Inkling as an open model is Mira Murati's team putting a stake in the ground against the frontier-lab consensus that one giant general-purpose model should serve every use case. The bet is that specialised, smaller models tuned for specific domains will beat generalist giants on cost and latency for enterprise deployment, which is exactly the argument enterprise buyers tired of GPT-5 and Gemini pricing want to hear. Thinking Machines still has to prove Inkling performs well enough on real workloads to justify switching costs, because open-weight competition from Meta's Llama line and Mistral means differentiation has to be substantive, not just philosophical. Enterprise AI buyers should treat this as validation that the market is fragmenting, and start testing smaller specialised models against their actual use cases rather than defaulting to the biggest name.