In the long, arduous road from drug discovery to commercialization, pharma companies pour millions of dollars into drug development with the full awareness that the drug being developed will probably fail.
That is the narrative that Brian Alexander, CEO of Valo Health — a Flagship Pioneering startup leveraging artificial intelligence and human data in drug discovery — is trying to flip. And the AI push notwithstanding, Valo’s story is less about efficiency and more about causality. As AI probes deeper within pharma with a whole host of companies declaring AI as a silver bullet to eliminate or reduce inefficiencies, this distinction is noteworthy.
“If you go pharma end-to-end, the inefficiencies of the end-to-end are in clinical development. So even if you save a few months up here — in the discovery phase — it doesn’t matter,” said Alexander in an interview in January. “How can you leverage efficiency, how can you leverage the ability to do things with a higher probability of success? You want a diversified portfolio, so we’re willing to take partial ownership of lots of drugs versus all ownership of one drug, because we’re changing the probability [of success] from 10 to 20%, not from like zero to 100%.”
This approach is intriguing, especially when it comes to new company formation, because conventional market and investor approaches have settled on companies having one or two ideas for which they raise money to fund clinical trials with the ambitious goal of future commercialization. But running clinical trials is not something that companies founded on scientific research and disease prevention are necessarily good at.
“My area of academic exercise is clinical trials, clinical development. I spent my whole life thinking about it,” Alexander said. “Why do people feel like they can just get some KOLs together and run a trial?”
So how is Valo looking at drug development powered by AI? Not in the lab. But in scouring human data and finding causal relationships.
“And so the way our human causal biology approach [works is that] there’s been some published research that shows that if you have a genetic basis for the target, it’s more than twice as likely to succeed when it gets into the clinic,” he explained. “And that’s not just genetics that are directly causal, but it’s using genetics and Mendelian randomization as kind of a causal instrument to understand the associations.”
Mendelian randomization (MR) is an epidemiological method that uses genetic variants as instrumental variables to examine the causal effect of a modifiable exposure — which could be a biomarker or a certain behavior — on an outcome, in this case, disease.
And Alexander’s hope is to be able to “industrialize that, even doubling that, say everything that we work on, now instead of having a 10% success rate when it goes to clinic becomes 20% — that would be transformational.”
The Lexington, Massachusetts company has access to more than 17 million de-identified patient records — some spanning 20 to 30 years — and biobank samples. The company aims to bring advanced AI to bear upon this data and uncover disease patterns, which in turn can identify potential targets for drug development. How Valo operates is that it works on multiple drug candidates and then partners with larger companies that have the wherewithal to commercialize the drug. A few partnerships have already been announced, including the three below:
- In November, the company announced that German drugmaker Merck was going to pay up to $3 billion to “leverage Valo’s AI-enabled human causal biology platform to identify and validate novel disease targets and its closed loop discovery platform to rapidly generate preclinical compounds” in Parkinson’s Disease and related disorders.
- In April 2025, Flagship Pioneering revealed that it was expanding its relationship with Pfizer to “discover compounds which may lead to the development of next-generation therapeutics for autoimmune diseases.”
- In January 2025, Novo Nordisk continued cementing its relationship, expanding its collaboration to discover and develop novel treatments in cardiometabolic disease.
But Valo Health is pursuing its own drug candidates as well, though one targeting diabetic retinopathy has already been paused after it failed to meet primary and secondary efficacy endpoints. But even if successful, Valo doesn’t intend to take any drug candidate to the finish line.
“We’re not going to commercialize anything ourselves,” Alexander said, pretty firmly. “So we don’t want to be forced to put all of our resources into pushing something into the clinic just so we have a clinical asset so we can do an IPO for a certain set of investors. We don’t want to let the investors drive the science.”
Photo: metamorworks, Getty Images
