A biotech company usually arrives with a precious object: one molecule, one platform, one piece of science that everyone agrees is the reason for the company. Arcus Biosciences arrived in 2015 with something stranger - a set of rules. Founders Terry Rosen and Juan Jaen wanted targets grounded in understandable biology, molecules designed to be differentiated, and enough control over the ingredients to test drug combinations without spending years negotiating for every bottle.
That logic came from experience. Rosen and Jaen had worked together for decades and previously built Flexus Biosciences, which Bristol Myers Squibb bought for $1.25 billion in 2015 to obtain an immuno-oncology program. At Arcus, they did not begin by carrying a leftover asset into a new corporate shell. They began, as the founders have put it, with a blank sheet of paper.
The result is a clinical-stage drug company in Hayward, California, with no approved medicine and therefore no product sales. It has, however, moved multiple internally discovered candidates into human studies, built one of biotech's more consequential alliances with Gilead Sciences, and advanced two programs into registrational Phase 3 trials. It also learned the hard lesson behind every portfolio: owning several shots does not make any one shot easy.
01 / The premiseA medicine cabinet, not a lottery ticket
Cancer adapts. A drug may block one escape route while a tumor finds another, which is why oncology increasingly relies on combinations. The mundane obstacle is that promising components often belong to different companies, each with its own priorities, data and deal terms. Arcus' answer was to build both small-molecule medicinal chemistry and antibody capabilities in-house, then design candidates with combinations in mind.
Its ARC studies use a platform design that can add or adjust cohorts under a shared clinical infrastructure. This matters less as a clever diagram than as a way to get answers. A candidate can be tested alone, paired with another Arcus molecule or combined with a partner's drug. Biomarker work then asks not only whether the tumor moved, but whether the intended biological mechanism moved with it.
“The real Arcus product is the ability to compose an experiment.”A useful way to read the company, not a company quotation
02 / The lead betCasdatifan moves to the center
Today, the center is casdatifan, an oral inhibitor of HIF-2alpha. That protein helps clear-cell renal cell carcinoma, the most common adult kidney-cancer subtype, respond to a disrupted oxygen-sensing system. The target is clinically validated: Merck's belzutifan is already approved in kidney cancer. Arcus is not inventing the category. It is trying to make a molecule with a competitive profile and then place it in combinations across several lines of care.
Early data are why the program has the floor. In a 100-milligram once-daily cohort of the Phase 1/1b ARC-20 study, Arcus reported a 45.2 percent confirmed response rate and median progression-free survival of 15.1 months at the January 30, 2026 data cutoff. Those numbers came from an early, non-randomized cohort. They are encouraging, not a substitute for a controlled Phase 3 result.
PEAK-1 is the decisive test now enrolling globally. It compares casdatifan plus cabozantinib with cabozantinib alone in people whose metastatic clear-cell kidney cancer has already been treated with immunotherapy. Arcus has also been exploring first-line combinations that could delay tyrosine kinase inhibitors, drugs that work but can bring a burdensome set of side effects. A separate Bristol Myers Squibb collaboration will test casdatifan alongside pumitamig, a PD-L1/VEGF-A bispecific antibody developed with BioNTech.
The subtle bit is the biomarker. Arcus researchers have studied suppression of erythropoietin, or EPO, as a signal that casdatifan is hitting HIF-2alpha. In July 2026, translational work from ARC-20 was published in Nature, connecting target engagement, tumor biology and clinical outcomes. A blood marker will not rescue an ineffective drug, but a credible one can help explain dose, response and patient differences instead of treating a trial as a black box.
03 / The second clockPancreatic cancer, then inflammation
The other late-stage candidate is quemliclustat, a small molecule that inhibits CD73. Tumors can use CD73 to generate adenosine, a chemical signal that suppresses immune activity in the tumor microenvironment. PRISM-1 is testing quemliclustat with gemcitabine and nab-paclitaxel against chemotherapy alone in previously untreated metastatic pancreatic ductal adenocarcinoma. Enrollment finished in September 2025, and the company expects results in the first half of 2027. The primary question is overall survival - a blunt endpoint, and the one that matters.
At the same time, Arcus has begun moving its discovery engine beyond cancer. AB102 is an oral antagonist of MRGPRX2, a receptor associated with mast-cell activation, with clinical development aimed at atopic dermatitis and chronic spontaneous urticaria. Behind it sit programs aimed at TNF, CCR6, STAT6 and CD40L, plus an anti-CD89 antibody. The commercial idea is legible: find disease areas dominated by injectable biologics and attempt to offer selective oral medicines. The scientific distance from a tidy preclinical profile to a useful prescription remains considerable.
If approved, Arcus drugs would be prescribed by oncologists and, later, specialists in inflammatory disease. Hospitals, clinics and insurers would be the commercial customers. Today the direct users are investigators and people volunteering for clinical trials.
04 / The resetWhat happened to TIGIT
Arcus' combination thesis once had a louder centerpiece: domvanalimab, an Fc-silent antibody against TIGIT, paired with its anti-PD-1 antibody zimberelimab. TIGIT became one of the pharmaceutical industry's crowded checkpoint targets. In April 2026, Arcus stopped the Phase 3 STAR-121 lung-cancer study after a planned futility analysis. The domvanalimab combination had not improved overall survival over pembrolizumab plus chemotherapy. EDGE-Lung was stopped with it, following earlier wind-down decisions in gastric cancer.
This is not a footnote. It is the cleanest evidence of the risk in Arcus' model. Rational biology and thoughtfully engineered molecules can still fail in a large randomized trial. The company responded by reducing related spending and placing more weight on casdatifan, quemliclustat and wholly owned inflammation assets. A broad pipeline became a more concentrated one.
The balance sheet gives that reset time. Arcus ended March 2026 with $876 million in cash, cash equivalents and marketable securities, and guided that its planned operations could be funded into at least the second half of 2028. It reported $17 million in first-quarter revenue, but this was collaboration accounting, not patients buying medicine. The same quarter produced a $128 million net loss. Clinical-stage biotech is a business in which cash is inventory and evidence is the hoped-for output.
05 / The businessGilead pays for a seat at the bench
Gilead is central to the economics. The companies signed a 10-year alliance in 2020 that delivered a $175 million upfront payment and a $200 million equity investment. In 2021, Gilead exercised options on three clinical-stage programs for $725 million. It later increased its ownership, including a $320 million investment announced in 2024.
For selected programs, the companies share global development costs, would split US profits and co-commercialize if a medicine is approved. Gilead generally controls commercialization outside the United States, subject to rights held by Taiho in parts of Asia, and would pay Arcus royalties. This gives Arcus capital and reach without turning it into a contract laboratory. It also creates dependence: priorities, reimbursements and development decisions are braided together with a much larger partner.
The competitive map changes by asset. In HIF-2alpha, casdatifan must distinguish itself from belzutifan and fit among well-established kidney-cancer combinations from companies including Merck, Bristol Myers Squibb and Exelixis. Quemliclustat must improve on a pancreatic-cancer chemotherapy backbone in a disease littered with failed experiments. In inflammation, Arcus is arriving after biologics built multibillion-dollar markets and oral immunomodulators established their convenience.
Arcus is not selling drugs yet. It is selling time, options and shared conviction to partners while the trials decide what the molecules are worth.
06 / The verdictWhere Arcus fits
Arcus sits in the middle tier of global biotech: too scientifically broad to be called a single-asset company, not yet a commercial pharmaceutical business, and large enough to run registrational programs while maintaining an internal discovery engine. Roughly 630 employees span discovery, clinical operations and the infrastructure required to work across partner-run and Arcus-run studies.
Its culture language is unusually practical. The company talks about camaraderie, urgency, integrity, simplicity and putting decisions with the people closest to the work. Its careers material promises a place where the rule book is still being written. That can mean autonomy; in biotech it also means the rule book is often revised by data.
What makes Arcus different is not a claim that it has found cancer's secret. It is the organizational choice to treat combinations as a design constraint from the beginning, supported by in-house chemistry, antibody work, flexible trials and pharmaceutical partnerships. The next proof points are plain: Phase 3 enrollment and data for casdatifan, the PRISM-1 survival result for quemliclustat, and the first human evidence from inflammation.
If those programs work, Arcus will look prescient for building a cabinet instead of guarding one bottle. If they do not, the cabinet will still be full of exquisitely reasoned experiments. Biology gets the final edit.