CH Health Tech Advisory

23 July 2026 · 1 min read

Why Tempus paid $1.5bn for Personalis: the data follows the patient

Tempus AI's $1.5B acquisition of Personalis, coming two months after the FDA approved atezolizumab with Signatera as companion diagnostic, looks to me like a bet on longitudinal patient data rather than current test volume—though I'm skeptical about how fast Personalis can replicate that regulatory path.

Tempus AI is paying $1.5 billion for Personalis, Inc., two months after the FDA approved atezolizumab for MRD-positive bladder cancer with Natera's Signatera as the companion diagnostic.

Once a ctDNA result decides who gets a drug, MRD has a defined commercial position. Tempus moved fast after that.

Natera ran about 250,000 clinical MRD tests last quarter. Personalis runs somewhere around ten thousand. At $1.5 billion, that price is buying something other than the current business.

My read is that they are buying the shape of the data. Most of what Tempus has assembled gives you one look at a patient: tumor sequencing at diagnosis, pathology since Paige. MRD repeats. Blood draws every few months for years, each one tied to the treatment given and what happened afterward. For a company building models on oncology data, a record that follows the same patient over time is a different input than a profile taken once.

When Tempus bought Paige I wrote that building a good product and finding a way to make money from it are separate problems, and that Paige had only solved the first. Personalis is further along on the second, with reimbursement moving and a category the FDA has now validated in at least one indication.

I'm less confident about the timeline. Signatera ended up in that label because Genentech built it into IMvigor011 years before the readout. Getting NeXT Personal to the same place means running trials of that size again, one indication at a time. That work is slow and expensive, and it runs on its own clock regardless of who owns the company.