Tempus

Precision Oncology Platform

Andrew Kuzemczak
May 13, 2025

Fact-based news without bias awaits. Make 1440 your choice today.

Overwhelmed by biased news? Cut through the clutter and get straight facts with your daily 1440 digest. From politics to sports, join millions who start their day informed.

Welcome to the 2 new subscribers who has joined Future Human since our last edition! Join 173 other leaders learning about the future of human health by subscribing here:

Hi friend,

Welcome back to Future Human! Last week, we explored the oldest startup yet (probably not a startup by most definitions), Headspace. I will not make it a normal thing to tackle a larger firm like that, but I still found the deep dive very exciting to craft. The data was copious given their extensive history, so research was smooth and opinions were abundant. I hope you enjoyed it as much as I did.

This week, I have prepared what I think is one of my favorite deep dives. Relax, I am not complementing myself as much as I am backing the startup. For context, here at Weill Cornell, we just finished the hem/onc unit (hematology/oncology, or blood and cancer). I have been on a cancer research bender for a few days, so diving into a startup in the space made sense. As I dove deeper into this one, the words flowed more easily and the future potential seemed endless. Again—not saying my writing is superb. Still mildly trash, but at least it’s evolving trash. The startup, however, potentially industry altering.

So with that tangent, let me ask you:

When a patient presents with cancer, what if their doctor could learn from every individual that came before them to predict what they will respond to best? If you are them, are you okay not trying everything, but just trying the one thing?

The Story

If you are reading Future Human, you have probably come across the concept of “precision medicine” before. I will not insult your intelligence by over explaining the concept (it is literally in the name), but for those wondering, it is just about the right treatment, for the right patient, at the right time. Doctors and researchers leverage data about a patient’s genetics, lifestyle, disease biology, and EHR (electronic health record) to predict which therapeutic will work the best. It is frequently touted as the future of medicine.

One serial entrepreneur with a history of industry-agnostic successes, has taken more steps than most to bring this dream to market. In 2015, after watching his wife receive a breast cancer diagnosis, Eric Lefkofsky founded Tempus AI to advance cancer diagnosis and treatment. It is his 14th company.¹

14, you say? Yeah, let’s get into it. He is the co-founder and General Partner of Lightbank, a private venture capital firm specializing in investments in technology companies. He has separately founded:

Brandon Apparel (1993) – clothing company acquired early in his career
Starbelly (1999) – internet company specializing in promotional products, sold to Halo Industries
BlueMedia, Inc. and Sea Blue Media LLC (1997, 2009) – media companies
InnerWorkings (2001) - print procurement services
Echo Global Logistics (2005) – transportation and logistics outsourcing firm
MediaBank (2006) - integrated media procurement technology
Groupon (2008) – global e-commerce marketplace offering daily deals
Pathos AI (2020) – AI-enabled drug development company

Who said you need to be a medical doctor, PhD, or MBA to be a healthtech entrepreneur? We have a lawyer breaking all the rules!

You may think that the household name of Groupon was his main success, but he has been stocking his coffers since the late 90’s. In 1999, Lefkofsky and his law school classmate Brad Keywell moved to Chicago to build Starbelly that was scaled—and sold—for $240 million in less than a year. But okay, how about an exit not during the frothy Dot-com bubble?

InnerWorkings went public after 5 years in 2006. Echo Global Logistics went public in 2009 (later acquired by PE firm, The Jordan Company). MediaBank would be acquired by Vista Equity Partners in 2014. I know. I know. PE may not be the ideal exit, but regardless, Eric made out with quite the sum of cash—he now applies that stockpile to Tempus.

Soon after founding the company in 2015, Eric brought on Ryan Fukushima, then an entrepreneur-in-residence and VP at Lightbank, to serve as COO of Tempus. With a small but growing team, they took to expanding. Compared to his previous 13 ventures, Tempus has taken a more measured path. Where as Groupon expanded into 50 countries in just three years, Tempus took nearly a decade to develop its foundation in the US. They just recently made a leap to build in Japan. Eric also dedicates all of his time to Tempus, which sharply contrasts his previous habit of jumping between multiple ventures. Finally, their 2024 IPO (nine years after founding) is much slower than his normal four to five year timeline. He is clearly applying a slower and more diligent philosophy to his biggest healthcare venture yet.²

“You have to convince me why I shouldn’t back this guy.”

Peter Barris, Chairman New Enterprise Associates

The Tech

The products Tempus offers are not the easiest to grasp upon first pass, so I will do my best to explain the suite. To begin, let’s go back to where it all started.

In the early years of the startup, their product was not too dissimilar from the backend of 23andMe (ideally, without the bankruptcy). They were not telling you where your ancestors were from, but they had a series of genomic sequencing tests and data monetization strategies that helped physicians learn more about you and pharmaceutical companies develop drugs based on your data:³^,⁴

Genomic sequencing – offers a broad range of DNA and RNA sequencing tests that physicians can order for their patients (provide insights on gene alterations associated with specific types of tumor)
Clinical data structuring – partners with health-care systems to collect, organize and aggregate clinical datas (natural language processing (NLP) transforms unstructured information into structured data)
Data monetization – sells curated and anonymous data to pharmaceutical companies for R&D of new drugs

So are they not selling your data now? Pffft, are you kidding? Of course they are. They are just doing a lot more as well. Here is a quick overview of what exactly Tempus does today.

Tempus aggregates real-world data from EHRs, pathology/radiology reports, genomic sequencing, and lab results. As mentioned above, the data are often fragmented, so their natural language processing (NLP) and AI pipeline cleans, extracts, and structures the information. Tempus also runs next-generation sequencing (NGS) on tumor tissue and assesses DNA mutations, RNA expression, rearrangements, and microsatellite instability. This data helps classify tumors more precisely and identify actionable targets for treatment. This next part blew my mind—the platform can estimate the origin of the tumor, assess how it will metabolize drugs, and assign an immunotherapy response score to inform oncologists of how well the patient will respond to each treatment. BOOM (mind blown). Then, Tempus trains machine learning models using its vast database of structured clinical/molecular data to predict which therapies might be most effective for a given patient. It can identify clinical trial matches in real time and provide oncologists personalized recommendations.⁵

Oncologists interact with the software in a number of ways:

Tempus Hub: where it all begins—platform for ordering tests, accessing patient data, and integrating AI-driven technologies into clinical workflows
Tempus One: AI-enabled clinical assistant that provides on-demand access to patient insights, streamlining the process of ordering and interpreting genomic tests
- Quickly access patient information, like report status and biomarker results
Tempus Trial Watch: program recommending clinical trials for which a patient may be eligible based on their profile
Tempus Next: care pathway platform that helps oncologists contextualize patients in their care journey and track patients for timely follow-up
- Track patients who need timely management and receive real-time insights into a hospital’s patient population
Tempus Pixel: AI-enabled insights from medical images
- Segment and measure lesions of interest, while linking them across time points for longitudinal tracking (this is where diagnostic radiologists start sweating)

Remember when I said at the start that the tech is a bit more challenging to grasp? What I just listed is really just the most front facing parts. That is the easy stuff. Below we have a bit more tech heavy features.

The algorithms
- IPS: immunotherapy score mentioned above—prognostic biomarker for immune checkpoint inhibitor candidates
- HRD: measures homologous recombination deficiency (cell's inability to repair DNA breaks—can lead to cancer)
- TO: refines diagnosis for cancers with uncertain origins
- DPYD: drug metabolism predictor mentioned above—identifies patients at elevated risk for toxicity to different chemotherapy
The patients
- Olivia: personal health concierge app brings patients the ability to consolidate their health data into one secure platform⁶

Tempus aims to become the everything app for both oncologists and their patients. From diagnosis to remission, all you should have to think about is Tempus. Although they are not there yet, the tech stack above leads me to believe that they are among the furthest along in this steep climb. We shall continue to watch closely.

The Market

The term “precision market” can encapsulate a large proportion of medicine. Although oncology dominates the “precision” conversation, it can also involve cardiology, neurology, and a few other disease areas. In the oncology space, think genetic tests to determine drug dosage, identifying patients at risk for specific cancers, and developing therapeutics based on tumor-specific mutations. Given all of that, the global precision medicine market size was estimated at $87.5 billion in 2023 and is projected to grow at a CAGR of 16.3% until 2030.

That massive size and complimentary massive growth is propelled by advances in genomics, increased demand for personalized treatments, and technological innovations in diagnostics. Additionally, rising prevalence of chronic diseases, favorable government initiatives, and increased investment in research (until recently) are likely to drive the market.⁷

A quick side note. I have been wanting to make this point since well before Future Human was launched. Notice how at the beginning I said the company was called “Tempus AI”. Since then, I have been calling it just Tempus. That is not just some silly attempt to minimize word count. Let me introduce you to the concept of AI washing.

I first heard this concept from Scott Galloway, so I cannot claim credit, but I still love it. Tempus was founded in 2015 as just that, “Tempus”. In 2024, however, with their IPO on the horizon, they renamed to “Tempus AI”. Given that AI was now half of their name, I bet you are thinking that it composes a large percentage of their business.

Yeah, I did to.

At the time of their rename, their IPO paperwork referred to to AI 228 times. Upon further research, we found that the AI Applications segment provided <2% of the company’s revenue.⁸

“I can’t decide if I should criticize Tempus or commend it. The market wants AI companies, Tempus wants the market’s capital, and it pairs the trade via AI washing.”

Scott Galloway

For me, these deep dives are always meant to be positive, yet honest. The AI labeling is a valuable detail to include, so I hope you chew on it.

More than half of Tempus revenue—55% in the first half of 2024—comes from genetic analysis. When a physician suspects a tumor, a biopsy sample is whisked away to a Tempus lab. The results are then cross-referenced with a 200-petabyte ML-trained dataset of similar cases. A report follows. The remaining 45% of revenue comes from licensing this extensive dataset to research institutions and major pharmaceutical companies. Companies can pay $150,000 to $500,000 per engagement, with multi-year deals with giants like AstraZeneca and GlaxoSmithKline coming in at $300 million each.

Okay, enough about Tempus alone, let us get back to the market.

With a market as big, profitable, and asset-light as precision medicine, you might venture to say that competitors are inevitable.

You might just be right!

Foundation Medicine⁹
- Founded in 2010 and now owned by Roche
- Offers a suite of gene tests to highlight clinically relevant biomarkers and alterations to help match patients to approved targeted therapies, immunotherapies, and clinical trial options
- FoundationOne CDX
  - FDA-approved tissue-based companion diagnostic for all solid tumors, indicated for 20+ targeted therapies
- FoundationOne Liquid CDX
  - FDA-approved blood-based companion diagnostic for all solid tumors, indicated for 8 targeted therapies
- FoundationOne Heme
  - Test for hematologic malignancies or solid tumors where novel gene fusion detection is desired
Flatiron Health¹⁰^,¹¹
- Founded in 2012 and became an independent affiliate of the Roche Group in 2018 (I think I know where Roche leadership has their sights set)
- Through clinical and data science, translate patient experiences into real-world evidence to improve treatment, inform policy, and advance research
- Flatiron Clinical Pipe™
  - Connects leading EHRs, enabling seamless, accurate data transfer, while accelerating study timelines, lowering data monitoring costs, and reducing burden on site teams
- Flatiron Assist™
  - Provides clinicians with treatment insights at the point of care via an EHR-integrated, intuitive decision-support tool
Guardant Health¹²
- Shield™ blood test
  - More pleasant and convenient screening option that helps overcome barriers associated with available methods
- Guardant360 TissueNext
  - Tissue biopsy test finds actionable information and improves PD-L1 (immune checkpoint) detection by >20% in non-small cell lung cancer
- Guardant Reveal
  - Tissue-free test that detects circulating tumor DNA (ctDNA) for minimal residual disease (MRD) assessment in early-stage colorectal, breast, and lung cancers

Alright, there you have it. Highly competitive market that everyone agrees needs investment and support. I predict as the economy stabilizes, interests rates fall, and venture capital returns to (at least something resembling) the 2021 free for all of monopoly money, doses, and mimosas, oncologic precision medicine is up first in healthtech to bathe in VC cash.

The Sick

Startups that are motivated by personal experience, like Tempus with Eric’s wife’s diagnosis, often end up becoming incredible teams to watch—all motivated by the purest passion to succeed. In Tempus’ case, the drive to better diagnose and treat cancer, is one nobody would argue is a challenge not worth pursuing.

Cancer can start almost anywhere in the human body, which is made up of trillions of cells. Tumors then can spread into, or invade, nearby tissues and can travel to distant places in the body to form new tumors (a process called metastasis). Cancer is also a genetic disease—that is, it is caused by changes to genes that control the way our cells function, especially how they grow and divide.

This is all to say, cancer is diverse. To best eradicate it, we cannot just waste time throwing the kitchen sink at each tumor. We must understand what cancer we are dealing with, as it could be one of thousands. A more precise way to label the patient’s cancer and link a treatment with the highest associated efficacy, would revolutionize not just the patient’s life, but also the life of the physician, care team, hospital administrator, and every individual involved in the cancer continuum.

In 2024, an estimated 2,000,000 new cases of cancer were diagnosed in the US. 612,000 people died from the disease. Prostate, lung, and colorectal cancers account for an estimated 48% of all cancers diagnosed in men, while the three most common cancers in women are breast, lung, and colorectal.¹³

As of January 2022, there were an estimated 18.1 million cancer survivors in the US. The number of cancer survivors is projected to increase to 22.5 million by 2032. These survivors received chemotherapy, hormone therapy, surgery, immunotherapy, radiation therapy, blood stem cell transplants, and targeted therapy. The data available from all of these stories is enormous. We know roughly what worked and what did not, when it worked and when it did not, and what mutations led to which outcomes. All we need is a team to locate, collect, and clean that information. Tempus and a bunch of AI may just be that answer.

As a newsletter that obsesses over the ‘future’, I also want to take a moment to acknowledge all the patients Tempus could also help beyond just cancer patients.

Cardiovascular disease is perhaps the only disease area that competes with cancer in terms of data available out in the ether. As mentioned countless times in Future Human, heart disease is the leading cause of death for men, women, and people of most racial and ethnic groups. One person dies every 33 seconds from cardiovascular disease, and one in 5 deaths are attributed to heart problems.¹⁴ Cardiology is a data rich environment with countless trends that can help inform physicians of best treatments to offer individual patients. In the spirit of data rich, there is no reason Tempus cannot then move into neurology and psychiatry with the amount of interactions recorded out there on the EHR for analysis. So much potential with a high speed team, endless data, cutting edge tech, and patients in need.

The Economy

Cancer care is financially devastating, most importantly for the patient and their family, but also for the country. The cost of cancer care in the United States, which was estimated at $183 billion in 2015, is projected to exceed $245 billion by 2030, an increase of more than 30 percent (cardiovascular disease care costs us ~$400 billion).¹⁵ As I wrote 3 weeks ago during the newsletter on Paige, a WHO study revealed treatment for cancer patients who have been diagnosed early are two to four times less expensive compared to treating people diagnosed with cancer at more advanced stages.

Cancer costs in the end-of-life stage vary considerably by cancer site, ranging from $71,300 for those with prostate cancer to $239,400 for those with acute myeloid leukemia (AML). Why the steep price?

You have probably heard the statement, a new drug takes 10 years and $1 billion to successfully make it to patients. Unlike most colloquialisms, this one carries a lot of truth. The cost to bring a cancer drug to market ranges from $650 million to $4.6 billion, with a median of $750 million.¹⁶

Tempus is targeting this economic burden and leveraging its TIME Network to support phase I clinical trials. The TIME network allows for rapid patient identification and site activation of trial sites. They have also launched their TIME Precision Network, a group of investigators leading the phase I study platform with a focus on activating and enrolling quickly across over 40 phase I-capable research centers.¹⁷ The cost of cancer drug development often lies in the immense time burden—AI-enabled patient identification could at least begin the process of shrinking this price tag.

Over the previous 60 years, the number of drugs approved in the US per billion dollars in R&D spending has halved every nine years. Half of that time and money is spent on clinical trials, which are growing larger and more complex. Still, only one in seven drugs that enters phase I trials is eventually approved. Although Tempus remains firmly focused on diagnosis and precision therapy, their AI could eventually accelerate most of the trial pipeline, from protocol writing to data analysis, cutting costs across the therapeutic journey.¹⁸

My Thoughts

The value of AI in drug development will take a bit longer to mature before it produces new therapeutics (think Isomorphic Labs from newsletter #7). With that said, AI armed with the treatments we already have and all of the patient data waiting to be analyzed can surely get current therapies to patients’ hands sooner for rapid, data-driven results. Tempus represents one of the purest play precision medicine startups on the market today. Driven by a desire to improve the experience Eric’s wife had, and equipped with billions of data points, this team appears unstoppable at the current moment.

Let us see all of the good they can do!

To more lives saved,

Andrew

I always appreciate feedback, questions, and conversation. Feel free to reach out on LinkedIn @andrewkuzemczak.