The Oncology Advantage: Why AI Could Change the Investment Math

Capital is following that progress, with cancer becoming a more compelling investment area. Researchers understand more of what drives the disease, can reach more of its vulnerabilities, and have better ways to measure whether a treatment is working.

One example: In May, Isomorphic Labs raised $2.1 billion, led by Thrive Capital, to expand its AI drug-design engine and advance programs toward human trials. That is a substantial bet on what better prediction could make possible. Its clinical results will determine how much of that promise holds up.^[2]

AI won’t cure cancer. But at AV, we believe that AI is especially well suited for applications in oncology as it accelerates the full biotech loop.

ABOUT THE AUTHORS

The Bottleneck Has Moved

Design tools and automated experiments are expanding the number of plausible drug candidates. However, the capacity to test them in people remains limited. Each trial needs funding, eligible patients, clinical sites, and an extensive safety record. A company can generate many ideas and still afford to advance only a few.^[2,3,4]

Even by utilizing digital twins and prediction-based models, the only way to fully understand whether drugs work or not is by testing them in people. That makes selection of which candidates to test increasingly valuable. Which cancer vulnerability matters? Which drug can reach it safely? Which patients are likely to respond? A wrong answer can consume years of work. A clearer answer before the next expensive trial can preserve capital and focus the company on a stronger program.

Better Biology Creates Better Starting Points

Cancer research has become more specific. Genetic testing can identify changes driving a tumor’s growth. New chemistry can reach proteins once considered exceptionally difficult to drug. Antibody-based medicines can carry a cancer-killing payload, while other drugs deliver radiation to tumors. These approaches give developers more ways to match a treatment to a defined problem.^[5,6,7]

Consider daraxonrasib, approved by the FDA in August 2026 for advanced pancreatic cancer. It targets RAS, a family of proteins that helps drive tumor growth. In a randomized trial of previously treated patients, median survival was 13.2 months with the drug versus 6.7 months with standard chemotherapy.^[5]

That progress gives investors something concrete to evaluate: a known biological problem, a medicine designed to address it, and evidence from patients. The strongest companies increasingly bring those pieces together.

Four Decisions Determine the Outcome

We organize the opportunity around four decisions: choose better, build better, test smarter, and learn faster. AI can contribute throughout the process, from ranking targets and screening potential safety problems to matching patients with trials and studying why a treatment stopped working.^[8,4]

The economic effect can show up before approval. A team that identifies an unsuitable candidate earlier can stop spending on it. A team that finds the right patient group can design a more focused test. Those improvements matter in a business where the cost of being wrong rises as development progresses.

AlphaFold’s advances in predicting protein structures and interactions illustrate the design opportunity. A clearer view of how molecules bind to proteins and how proteins fold and fit together can help researchers decide what to build and test. Experimental results remain essential, especially when predictions move from a computer model into a patient.^[8]

Cancer Gives AI More to Learn From

Oncology combines genetic information, tissue images, scans, treatment histories, and patient outcomes. Project GENIE’s 2025 public registry included more than 211,000 patients and 250,000 tumors. The Cancer Genome Atlas characterized more than 20,000 tumor and matched-normal samples across 33 cancer types.^[9,10]

The useful connection is between biology, treatment, and outcome. A tumor carries a particular alteration. A patient receives a drug. The cancer responds, returns or develops resistance. In some settings, scans and blood tests provide early feedback; durable benefit still requires follow-up. We think this combination makes oncology a particularly promising proving ground for AI.

Data quality creates its own risk. Records can be incomplete, patient populations differ, and a pattern in past results may disappear in a new trial. The useful test is whether a model improves decisions on patients or experiments it has never seen.^[11]

For a company, the advantage grows when its own experiments and clinical results improve the next decision. Access to reliable, treatment-linked data can be difficult to reproduce. That deserves close attention in diligence.

The Paper Trail Is Part of the Opportunity

Drug development also involves a large amount of document work. Teams assemble safety findings, manufacturing records, trial plans, and responses to regulators. Ruxandra Teslo’s (genomics PhD and biotech writer) recent New York Times essay describes how cost and administrative delays can constrain the ability to test promising medicines.^[12]

AI can help teams find relevant evidence, compare public FDA decisions with their own records, spot inconsistencies, and prepare better questions for regulatory meetings. AI could even become a part of the entire process to make the regulatory work more streamlined and automated. Imagine if the FDA had an operating system that quickly mapped findings from submissions, highlighted missing pieces, and responded faster by learning from every submission.

The investment value could be practical. Less repeat work, fewer avoidable delays, and more time spent resolving the questions that matter. Every output still needs traceable sources and expert review. Patient safety and regulatory accountability remain essential.

The Evidence Is Starting to Reach Patients

Isomorphic is one expression of this opportunity. Another is Insilico Medicine, which connects AI-assisted target selection and drug design with clinical development. As of June 2026, it reported 31 selected candidates, 13 clearances for human testing, and 10 clinical programs across several diseases.^[13] There is still no AI-designed approved drug from them, but they are showing how AI can make the entire process faster and more streamlined. Hopefully some of their candidates prove effective in the clinic soon.

Formation Bio is another example of AI-native company. It focuses further downstream, using AI to help select repurpose, and develop existing drugs. It reported moving one program from licensing to its first human dose in five months. That example was outside oncology, and the announcement does not isolate how much of the speed came from AI, but they are certainly being very efficient with their drug development work.^[14]

These are meaningful steps. They also leave a major question open: whether AI will produce a sustained improvement in clinical success and approval rates. The evidence remains early. We would look for measured improvements in a company’s development decisions before assigning value to the broader promise.^[11]

Buyers Are Paying for a Clear Advantage

The buyer side of the market is already visible. Pfizer’s Seagen acquisition carried an enterprise value of about $43 billion. Gilead agreed to pay $3.15 billion upfront for Tubulis, whose targeted antibody drugs had entered clinical testing. GSK agreed to pay $1 billion upfront for IDRx, built around a precision drug for a defined cancer population.^[6,15,16]

Our reading is that buyers value differentiated drugs, credible human evidence, and capabilities that would be difficult to build themselves. Exceptional preclinical programs can also command substantial prices, as Novartis’s $1 billion upfront agreement for Mariana Oncology illustrates.^[17]

We favor companies with clear biology, a realistic path to informative human data, and a reason to win against competing treatments. AI is valuable when it strengthens that case. Entry price, financing needs, and the quality of the lead investors still determine whether the opportunity is attractive.

Patient-selection tests and recurrence monitoring also interest us when the result changes a treatment decision. A company needs to show how its information will be used, who benefits, and who will pay for it.

Six Examples From the AV Portfolio

AV has backed companies addressing different parts of this process. Together, these examples show the breadth of the science and development work behind the thesis.

The Risks Still Require Judgment

Tumors evolve. Side effects can limit an otherwise promising medicine. Strong animal results can fail in people, and a poorly designed trial might leave the central question unanswered. AI can compound these problems when it learns from biased data or produces confident errors.^[11,19]

A growing market can also attract too many similar companies at inflated prices. The work remains selective: strong science, a capable team, enough capital to reach a meaningful result, and disciplined terms.

The Time to Build Exposure Is Now

Better cancer biology, richer data and more capable development tools are converging while many of the companies using them are still private. We believe this is a time to build selective exposure. The opportunity is to recognize which teams can turn those advantages into useful human evidence before that progress is fully reflected in the price.

Through Alumni Ventures, accredited investors can access selected biotech opportunities alongside leading venture investors experienced in the sector, with diligence from a team evaluating the science and investment terms. Explore AV’s healthcare and life-sciences opportunities to see how to participate.

Explore AV healthcare and life-sciences opportunities ->

Venture capital investing involves substantial risk, including risk of loss of all capital invested. Achievement of investment objectives cannot be guaranteed.

Sources

1. American Cancer Society. Cancer Statistics 2026: mortality and projected U.S. cancer burden. January 13, 2026.
2. Isomorphic Labs. $2.1 billion Series B funding announcement. May 12, 2026. Company-reported.
3. Minikel et al., Nature. Refining the impact of genetic evidence on clinical success. April 17, 2024.
4. U.S. FDA. Artificial Intelligence for Drug Development. Includes experience with 500-plus submissions in 2016-2023.
5. U.S. FDA. Daraxonrasib approval and randomized pancreatic-cancer trial results. August 26, 2026.
6. Pfizer. Completion of Seagen acquisition. December 14, 2023.
7. AstraZeneca. Agreement to acquire Fusion Pharmaceuticals. March 19, 2024.
8. Abramson et al., Nature. Accurate structure prediction of biomolecular interactions with AlphaFold 3. May 8, 2024.
9. American Association for Cancer Research. Project GENIE, 2025 annual report. Patient and tumor counts refer to the cited 2025 release.
10. National Cancer Institute. The Cancer Genome Atlas Program: molecular characterization and sample scope.
11. Bender et al., Nature Reviews Drug Discovery. Artificial intelligence in drug discovery: what it is, where we stand and the path forward. August 7, 2026.
12. Ruxandra Teslo, The New York Times. The Search for New Cures Is Broken. September 4, 2026. Guest essay; cited as commentary.
13. Insilico Medicine. First-half 2026 update. July 9, 2026. Pipeline counts are company-reported as of June 30, 2026.
14. Formation Bio. First participant dosed in the BLKR201 trial. June 9, 2026. Company-reported execution milestone.
15. Gilead Sciences. Agreement to acquire Tubulis. April 7, 2026.
16. GSK. Agreement to acquire IDRx. January 13, 2025.
17. Novartis. Agreement to acquire Mariana Oncology. May 2, 2024.
18. Bristol Myers Squibb. Agreement to acquire RayzeBio. December 26, 2023.
19. U.S. FDA. Withdrawn Cancer Accelerated Approvals.

This communication is from Alumni Ventures, a for-profit venture capital company that is not affiliated with or endorsed by any school. It is not personalized advice, and AV only provides advice to its client funds. This communication is neither an offer to sell, nor a solicitation of an offer to purchase, any security. Such offers are made only pursuant to the formal offering documents for the fund(s) concerned, and describe significant risks and other material information that should be carefully considered before investing. For additional information, please see here. Achievement of investment objectives, including any amount of investment return, cannot be guaranteed. Co-investors are shown for illustrative purposes only, do not reflect all organizations with which AV co-invests, and do not necessarily indicate future co-investors. Example portfolio companies shown are not available to future investors, except potentially in the case of follow-on investments. Venture capital investing involves substantial risk, including risk of loss of all capital invested. Diversification cannot prevent investment loss; it is a strategy to mitigate investment risk. This communication includes forward-looking statements, generally consisting of any statement pertaining to any issue other than historical fact, including without limitation predictions, financial projections, the anticipated results of the execution of any plan or strategy, the expectation or belief of the speaker, or other events or circumstances to exist in the future. Forward-looking statements are not representations of actual fact, depend on certain assumptions that may not be realized, and are not guaranteed to occur. Any forward-looking statements included in this communication speak only as of the date of the communication. AV and its affiliates disclaim any obligation to update, amend, or alter such forward-looking statements, whether due to subsequent events, new information, or otherwise.