Recently, one healthcare AI company suddenly came into the spotlight. It seems to meet all the right conditions—and when Nancy Pelosi, often dubbed the “Goddess of Stocks on Capitol Hill,” openly spoke about this stock earlier this year, it sparked a frenzy among investors. Yes, it’s Tempus AI.
If we ask where AI can create the most value, healthcare is undoubtedly one of the top contenders. It’s one of the most valuable industries in human society. In the U.S. alone, healthcare spending accounts for 20% of the nation’s GDP annually. Yet, the healthcare industry is facing many challenges: doctor shortages, high misdiagnosis rates, and ever-rising costs. With an aging population, these problems will only become more serious. Against this backdrop, the integration of AI into healthcare is inevitable. Healthcare AI is poised to be a massive opportunity, and any company that can establish itself in this space could become immensely valuable in the future. So, in the following, we shall have a deeper look into the company, and how it differentiates itself from its competitors.
The Origin of Tem
Tempus AI was founded by Eric Lefkofsky. Eric is a well-known entrepreneur in Silicon Valley, having built five companies before this—most notably, Groupon. He was already a billionaire. But in 2014, his life was shaken when his wife was diagnosed with breast cancer.
During the treatment process, he was shocked to discover how little most doctors actually knew about their patients. Many couldn’t even access full patient histories and were often relying purely on experience to make diagnoses. As someone familiar with the transformative power of big data and AI in other industries, Eric found it unacceptable that healthcare—where lives are at stake— was so technologically backward.
He believed that the healthcare industry needed a deep reform, one where technology empowered doctors to make more accurate decisions based on comprehensive data.
The Data Moat
At the start, Eric and his team built a medical data platform that could clean, integrate, and standardize fragmented data, then extract valuable insights using AI. But the real challenge wasn’t building the platform—it was acquiring the data. Medical data is highly fragmented across hospitals, labs, and research institutions, with inconsistent formats and even handwritten records that can’t be used in AI analysis.
To tackle the problem, first, Tempus offered low-cost genetic sequencing services to attract patients and gather data. But that was just the beginning. They spent years negotiating with hospitals and institutions one by one, collecting data ranging from treatment outcomes to physician notes and test results—of course, all anonymized and privacy-protected. Much of this data still had to be manually sorted and standardized.
While progress was slow in the early years and commercialization lagged, these tedious and labor-intensive efforts laid a strong foundation. After nearly a decade, Tempus has built partnerships with over 1,000 hospitals, 65% of academic medical centers in the U.S., and 50% of oncologists. This built the data moat for Tempus.
In the pharmaceutical world, Tempus has also made strides. It has over 200 biopharma partners, including 19 of the top 20 U.S. pharma companies. Today, Tempus boasts one of the largest medical data lakes in the world—with 240 PB of data, 9 million clinical records, and 1.2 million imaging records—more than any public database, and 50x larger than the U.S. government’s TCGA cancer database.
This includes 220,000 records that contain matched clinical, imaging, genomic, and transcriptomic data real-world data with high quality.
In fact, Tempus’s long-term vision is precision medicine. Traditional medicine treats most patients the same way for the same condition, but people differ in genetics, environment, and lifestyle. Precision medicine uses big data and AI to tailor treatment to individual characteristics.
AI can analyze your data, compare it to global patient profiles, and recommend optimal treatment something human doctors simply can’t match. Eric has said that in the next 10 years, AI will be a doctor’s assistant, but the long-term goal—30 to 50 years—is for AI to become a more reliable decision-maker than humans.
If realized, this vision could save enormous costs and lives—value that would eventually be reflected in the company’s valuation.
Business Model
But how will Tempus get there? Currently, it has two main revenue streams:
One, is Genetic Sequencing – not only the original business but still the core revenue driver (65%). They don’t just test, they analyze results using AI to recommend better treatments—like identifying key mutations in cancer patients and suggesting targeted drugs. In 2024, they performed over 270,000 tests, with each test costing $1,530 (mostly covered by insurance).
Second, is Data & Services – the other 35% comes from working with pharma companies. When developing new drugs, it’s hard to find suitable trial participants. Tempus uses its database to help find qualified patients quickly, cutting recruitment time from 8 months to 14 days, improving trial success rates. They also help design smarter clinical trials.
There’s also a third, smaller business line: AI Applications, such as their diagnostic platform “Next.” This seems like their most “AI-looking” product, but it currently contributes little to revenue.
Tempus' Strategy: Data First, Profit Later
Both sequencing and pharma services might seem traditional and unglamorous, but they’re all about one goal: collecting high-quality data. Eric has said so directly—their current goal isn’t profitability but maximizing data acquisition. Whether from sequencing, partnerships, or even consumer apps like their new “Olivia” health-tracking app, it’s all about gathering more data.
In fact, their latest acquisition of Ambry (a sequencing company in pediatric, rare disease, and cardiology) was clearly to expand their data reach beyond cancer.
Financially, Tempus is not profitable and isn’t trying to be—for now. Operating losses are about $700 million per quarter, more than their total revenue. They’re funding operations with investor capital, not organic cash flow.
Still, this makes sense within their strategy. Like Tesla, the more users they have, the better their AI becomes. Better AI attracts more users, creating a positive feedback loop. In this system, data is the key moat.
Conclusion
In fact, Tempus indeed looks like a textbook example of a “good company”: Having a massive addressable market, strong competitive moat (data), experienced and visionary founder, as an early leader in an emerging AI megatrend.
But a good company ≠ good investment.
Despite strong revenue growth (30–45%), the company is deeply unprofitable, with negative cash flow and high debt (50% long-term debt ratio). Even optimistic forecasts don’t expect positive
cash flow until 2026–2027. Any slowdown in AI or data acquisition could force additional fundraising and hurt
shareholders.
Moreover, the company has no clear path to near-term profitability. Management is focused on the long-term vision, not current earnings. This deters institutional investors who need near-term certainty.
But overall speaking, Tempus is definitely a company that is worth acknowledging.
