Exelixis enters the back half of 2026 with two questions hanging over the same stock. The first is a regulatory event with a named calendar date. The FDA target action date for the first New Drug Application of zanzalintinib, an oral kinase inhibitor in previously treated metastatic colorectal cancer, is December 3, 2026. The second is a legal one that just moved: late in August the Federal Circuit affirmed the district court judgment that three cabozantinib patents are not invalid, holding the earliest possible effective date for the MSN generic application in early 2030. The market has already responded to the second. The shares sit near $59, up from a fall low in recent memory, at a market capitalization close to $14.6 billion.
The franchise that funds everything is cabozantinib, sold in the United States as CABOMETYX and COMETRIQ and marketed by partners outside the country. Second-quarter total revenues reached $628.7 million, up 11% year over year. GAAP diluted earnings per share stood at $0.82. Full-year guidance sits at $2.55 billion in total revenues, trimmed by roughly $75 million from the prior August guidance on slower patient kinetics in the neuroendocrine tumor indication.
Three variables carry the thesis from here. The first is the December PDUFA outcome for zanzalintinib and the commercial ramp that follows in colorectal cancer. The second is the STELLAR-304 top-line readout for zanzalintinib plus nivolumab in non-clear cell renal cell carcinoma, expected in the second half of 2026. The third is the durability of cabozantinib volume in a second line that is increasingly crowded, with the first generic approval no earlier than January 15, 2030 but three separate applications already in the pipeline. The falsifiable test is the December 3, 2026 action date itself.
The counterargument deserves a sentence of its own before the sections below. The stock has roughly doubled off its low and now embeds a fair amount of zanzalintinib value, while the guidance cut signals that even the legacy franchise is growing slower than a year ago. The debate is no longer whether cabozantinib is a durable drug. It is how much the market is already paying for the second generation of the pipeline before it has earned the label.
Exelixis, Inc. is a Delaware corporation headquartered in Alameda, California, listed on Nasdaq under the ticker EXEL and registered with the SEC as a domestic large accelerated filer. The company reports under U.S. GAAP on a fiscal year that ends near the last Friday of December. The operating model is that of a late-stage oncology company with one dominant commercial franchise, a lead clinical asset under active FDA review, and a broader early-stage pipeline that spans small molecules and biotherapeutics.
The commercial center of gravity is the cabozantinib franchise, the product that has carried the company since its first approval in 2016. Cabozantinib is an inhibitor of multiple tyrosine kinases, including MET, AXL, VEGF receptors and RET, and it is approved in the United States as CABOMETYX tablets for several indications and as COMETRIQ capsules for metastatic medullary thyroid cancer. Outside the United States and Japan, Ipsen develops and commercializes the franchise, and in Japan Takeda holds those rights. The structure converts a single molecule into three revenue streams: U.S. net product revenues recognized directly by the company, royalties and milestone payments from Ipsen and Takeda, and collaboration services revenue that includes development cost reimbursements.
The strategic pivot is the second-generation kinase program built around zanzalintinib, an oral inhibitor of kinases including the TAM kinases, MET and VEGF receptors. The development program now includes three pivotal trials, a series of phase 1b/2 and phase 2 expansion cohorts in new indications, and two partnership-sponsored studies with Merck. The company frames the strategy as building next-generation oncology franchises, and it has expanded its U.S. gastrointestinal sales team in 2026 to accelerate cabozantinib momentum in neuroendocrine tumors while preparing the commercial infrastructure for a potential zanzalintinib launch.
Two other commercial relationships round out the portfolio. COTELLIC, a MEK inhibitor, is marketed in collaboration with Genentech for specific forms of advanced melanoma, and MINNEBRO, an oral selective mineralocorticoreceptor blocker for hypertension, is approved in Japan and licensed to Daiichi Sankyo. Both contribute at a level that matters to the income statement mix but not to the direction of the stock. The capital allocation program has been aggressive. The company has repurchased roughly $2.9 billion of stock over the past three years, retiring nearly 94 million shares as of the most recent quarter. It completed its prior repurchase program last spring before the Board authorized a fresh $750 million program running into next year.
The first moat is the cabozantinib patent estate, and its strength is best understood as a wall of litigation rather than a single filing. The Orange Book lists nine patents for CABOMETYX, and the company has litigated or is litigating every one of them against MSN, Azurity, Handa, Accord and Almatica. The Federal Circuit affirmation in late August 2026 of three cabozantinib patents, each of which expires in early 2030, is the most consequential ruling of the year. It confirms that the MSN generic tablet application cannot take effect earlier than that date. A separate patent trial, set for early November 2026, covers the final patent in the estate, which runs into 2032. The company also holds regulatory exclusivity layers on top of the patents.
The second moat is the franchise's embedded position in the renal cell carcinoma treatment algorithm. CABOMETYX with nivolumab is a first-line combination in advanced renal cell carcinoma, the setting with the largest patient population, and refills on that combination are a major driver of the 6% unit growth reported in the first half of 2026. The company reported that CABOMETYX remains the most prescribed TKI monotherapy and the most prescribed TKI plus immune checkpoint inhibitor combination in first-line advanced renal cell carcinoma. The neuroendocrine tumor indication, approved in March 2025, adds a second high-value population, though patient kinetics in that indication proved slower than the prior guidance assumed, which is the specific reason the full-year outlook was trimmed in August.
The third moat is the zanzalintinib program itself, and it is moat in the sense of clinical credibility rather than legal protection. The STELLAR-303 pivotal trial in previously treated metastatic colorectal cancer demonstrated a statistically significant overall survival improvement versus regorafenib. The final analysis showed a 20% reduction in the risk of death, with a hazard ratio of 0.80. The results were presented at the 2025 ESMO Congress and published in The Lancet. A second endpoint, overall survival in the subset of patients without active liver metastases, trended in the same direction at a hazard ratio of 0.83 but did not reach statistical significance, a nuance the market is likely to weigh when the NDA is reviewed. The STELLAR-304 trial in non-clear cell renal cell carcinoma and the STELLAR-311 trial in neuroendocrine tumors extend the platform into two additional large oncology populations.
The early-stage pipeline is the fourth and weakest layer. Four programs are in phase 1 or earlier and are described in the filings as having best-in-class potential: XL309, a USP1 inhibitor for BRCA-mutated tumors; XB010, an antibody-drug conjugate targeting the 5T4 tumor antigen; XB628, a first-in-class bispecific antibody targeting PD-L1 and NKG2A; and XB371, a tissue factor-targeting antibody-drug conjugate. None of these has de-risked enough to contribute to a valuation. The honest framing is that the moat is a franchise moat built on one molecule and its patent wall, with a second molecule that has earned clinical credibility but not yet commercial validation.
The second-quarter income statement is the clearest expression of the company's current state. Total revenues reached $628.7 million, up double digits year over year. Net product revenues, almost entirely cabozantinib, grew at a similar pace, and collaboration revenues from partners rose on higher royalty inflows. GAAP net income was $212.1 million, or $0.82 per diluted share, with non-GAAP diluted earnings per share slightly higher after excluding stock-based compensation and its tax effect. Gross margin held near the top of the range, a figure that reflects both the high-margin profile of oncology products and the royalty payable on U.S. cabozantinib sales. The first half shows the same pattern in aggregate, with total revenues above $1.2 billion and GAAP net income above $400 million for the period. Operating cash flow was the standout line, nearly double the prior-year figure, on the back of higher revenues and a working capital tailwind. That cash generation is what funded the $753.3 million of share repurchases in the first half, with the second quarter alone accounting for roughly half of that at an average price in the high $40s per share.
The expense profile tells a second story. Research and development spending for the half was $411.9 million, broadly flat year over year, with clinical trial costs shifting toward zanzalintinib, XB628 and XB371 as cabozantinib and XL309 costs declined. Selling, general and administrative expenses were $287.2 million, up 6%, reflecting the expanded U.S. gastrointestinal sales team and early commercial preparation for a potential zanzalintinib launch. The income statement is running at roughly a 40% operating margin on total revenues, a level of profitability that is distinctive for a company with a single dominant product and that is the reason the guidance cut in August landed as a meaningful event rather than a routine reset.
The balance sheet is strong and is moving in the direction the company intends. Cash, cash equivalents and marketable securities stood at $1.38 billion at the end of the half, down from the start of the year on the pace of repurchases, and working capital was $888.9 million. There is no meaningful debt load, and the accumulated deficit, which had narrowed considerably in recent years, widened modestly over the half on equity activity. Full-year guidance, updated in early August, is $2.55 billion in total revenues, trimmed from the prior plan, and $2.35 billion in net product revenues, also trimmed. The guidance explicitly excludes any revenue from a potential U.S. approval and launch of zanzalintinib, which means a favorable December outcome is pure upside against the current plan. Research and development expense guidance was trimmed, and the assumed effective tax rate sits in the low 20s.
The single most important financial dynamic is the interaction between the cash flow machine and the buyback program. The company has retired nearly 94 million shares at an average price in the low $30s per share, and nearly $600 million remains available under the current authorization running into next year. That combination, a growing earnings base plus a shrinking share count, is doing much of the work in the per-share numbers, and it explains why non-GAAP earnings per share can rise faster than revenue even in a quarter where guidance was cut. The risk is that the buyback is funding itself from a franchise whose growth rate is beginning to slow, and the August guidance cut is the first visible evidence of that tension.
The next two quarters carry the entire forward story, and the calendar is dense. The first event is the FDA decision on the zanzalintinib NDA, with a target action date of December 3, 2026, for the colorectal cancer indication. An approval converts the lead asset from a clinical value driver into a revenue line item and begins a commercial ramp that the U.S. gastrointestinal sales team has been building specifically to absorb. A rejection, or a complete response letter that pushes the decision well past the target date, is the single largest negative catalyst available in the current price, because the stock has already moved meaningfully off its low on the assumption that the approval is probable.
The second event is the STELLAR-304 top-line readout in the second half of 2026, the pivotal trial of zanzalintinib plus nivolumab versus sunitinib in previously untreated non-clear cell renal cell carcinoma. Non-clear cell renal cell carcinoma is a population with limited approved options and with biology that is plausibly sensitive to MET and VEGF inhibition, which is the mechanistic rationale for the combination. A positive readout extends the zanzalintinib franchise into a first-line renal cell cancer setting and directly supports the pipeline story that the market has already begun to price. A miss would isolate the approval case to colorectal cancer and change the valuation math accordingly.
The third event is the early November trial in the consolidated patent litigation against MSN, which carries the final patent in the estate, set to run into 2032. The August Federal Circuit ruling already protected the earlier date, but the trial outcome determines whether the 2030 boundary is the real one or whether it extends toward 2032. The company also faces two additional applications that have not yet been litigated to conclusion. One, from Handa, received tentative FDA approval in late July, and the other, from Almatica, is still being evaluated by the company's legal team. Each of these represents a separate path to a cabozantinib product reaching the market, and the timing of each depends on court and agency outcomes that are outside the company's control.
The execution risks are concentrated in three places. The first is the commercial ramp of zanzalintinib, which requires the company to build a new U.S. launch machine on top of a franchise that is still growing. The second is the durability of cabozantinib volume in neuroendocrine tumors, where the August guidance cut was specifically attributed to slower patient kinetics, and in first-line renal cell carcinoma, where the combination with nivolumab faces an increasingly competitive field. The third is the concentration of the early-stage pipeline, which the company expects to advance by progressing up to two new development candidates into preclinical development this year, an output level that is meaningful for the pipeline but not for the near-term income statement. The regulatory and policy environment adds a second layer of execution risk that applies to every product in the portfolio. The Inflation Reduction Act's Medicare drug price negotiation program is the principal one, and the company has qualified for the small biotech exception with respect to the cabozantinib franchise through the 2028 initial price applicability year, with the intention of maintaining that status and the associated price floor for several more years. The Part D manufacturer discount program, which raises the share of cost borne by manufacturers, has a multi-year phase-in for the company on specified small manufacturer status. Beyond those, the administration's most favored nation pricing pilots and the spring proclamation imposing tariffs of up to 100% on certain imported patented pharmaceuticals remain open policy questions whose timing and scope are uncertain. None of these is a near-term threat to the current plan, and all of them are relevant to the long-dated cash flows the valuation section assigns to the franchise. The counterpoint to the forward case is that much of the good news is already in the price. The Federal Circuit win, the NDA acceptance, and the positive data from the pivotal colorectal cancer trial all preceded the current level, and the stock has roughly doubled off its fall low. What remains to be earned is the approval itself, the non-clear cell readout, and the evidence that cabozantinib volume can keep growing in a slowing market. The gap between the two lists is the entire forward risk profile, and it is narrower than the price implies to anyone who has been holding since the low.
The largest single risk is a negative or delayed regulatory outcome for zanzalintinib in the colorectal cancer indication. The NDA is under active standard review with a December 3, 2026 target action date, and the company's own disclosure flags the inherent uncertainties of the review process. A complete response letter is the most likely adverse intermediate outcome, and it would not end the program but would push the commercial start back by a period that depends on the FDA's requirements, typically on the order of a year. The valuation impact of a slip is difficult to size precisely, but it is the scenario the current price has the least margin for, because the stock's recovery off its low has been substantially driven by the expectation of an on-schedule approval.
The second risk is the cabozantinib generic wall, and it is a multi-party event rather than a single one. The Federal Circuit ruling in late August holds the MSN tablet application at an early-2030 earliest effective date, but the final patent trial is set for early November and could extend that boundary toward 2032. The Handa application received tentative FDA approval in late July, a development that compresses the timeline for at least one competitor, and Almatica's application is still being evaluated. The realistic bear case is not a single generic entering in 2030 but a sequence of approvals and litigations that erodes the U.S. franchise over 2030 through 2032, with the net effect on cash flow depending on how quickly each entrant captures share and how the pricing of the U.S. market adjusts in response.
The third risk is the competitive and pricing environment for the cabozantinib franchise in the near term. The neuroendocrine tumor indication is the specific source of the August guidance cut, and the company's own disclosure attributes it to slower patient kinetics, which is a demand-side problem rather than a supply or execution one. First-line renal cell carcinoma, the largest population, is a market with multiple active combinations, and the 6% unit growth reported in the first half is a growth rate, not a momentum rate, for a product in a maturing indication. The policy layer, Medicare price negotiation exposure after the small biotech exception sunsets, Part D discount phase-in, and the open most favored nation and tariff questions, sits on top of both.
The downside scenario that combines these risks is a year in which the NDA slips, STELLAR-304 disappoints or is delayed, and cabozantinib volume growth slows further. In that scenario the earnings base that funds the buyback program grows more slowly than the share count declines, and the per-share math that has powered the multiple begins to stall. The scenario that is less likely but more severe is an adverse '039 patent trial outcome in November 2026 combined with a Handa approval that takes effect quickly, which would pull the front of the generic wall forward and change the long-dated cash flow model for the franchise.
The valuation frame starts from the earnings the company actually produces. The full-year guidance midpoint is $2.55 billion in total revenues, with $2.35 billion in net product revenues, and expense guidance that implies an operating margin in the high 30s to low 40s on total revenues, before tax. Against a market capitalization of roughly $14.6 billion and about a quarter billion shares outstanding, the stock prices at a multiple on current-year net income that is in the mid-to-high teens on a GAAP basis and lower on a non-GAAP basis, once the tax effect of stock-based compensation is included. The cleanest anchor is that the market is paying roughly $14.6 for every dollar of 2026 revenue on the guided base, before any contribution from zanzalintinib.
The three-scenario frame runs from the bear to the bull on two variables, the 2026 earnings trajectory and the value assigned to zanzalintinib. The bear case holds the cabozantinib franchise roughly flat to slightly declining in volume growth, prices in a full year or more of NDA delay, and assigns the zanzalintinib program a discounted value reflecting the delay and the STELLAR-304 uncertainty. In that frame the stock's support comes from the buyback program and the cash flow, and the implied multiple compresses toward the low end of the current range. The base case assumes the December approval arrives on schedule, STELLAR-304 is positive, cabozantinib volume grows in the low single digits, and zanzalintinib begins a modest commercial ramp in 2027 that the current guidance does not yet capture. The bull case adds a strong readout in the non-clear cell trial, a faster-than-expected zanzalintinib ramp, and a favorable outcome in the final patent litigation that extends the generic boundary toward 2032, which raises the present value of the long-dated franchise cash flows.
The buyback program changes the multiple arithmetic in a way that is worth isolating. With nearly $600 million of authorization remaining and a cash flow base that generated well over $500 million in operating cash in the first half alone, the company can retire a further high single digit percentage of the share count before the end of next year without touching the balance sheet. That is the difference between a multiple that is static and one that mechanically declines on a per-share basis even if the enterprise value is flat. It is also the main reason the bear case has a floor: the cash flow, even in a flat franchise scenario, is large enough to fund a continued reduction in share count at a price the Board has shown a willingness to pay.
The honest constraint on the valuation is that the market is paying for a franchise whose first drug has a 2030 generic boundary and a second drug that has not yet been approved. The bear case says the price is right for the base case and rich for the bear. The bull case says the price is cheap for what the pipeline is producing if the next two catalysts land. The base case, an on-schedule approval and a solid STELLAR-304, is the scenario the current multiple is built on, and it is the one that leaves the least room for either a pleasant surprise or a painful miss.
Exelixis is a company in the second act of a franchise, and the assessment has to be framed around what the current price already contains. The late-August Federal Circuit ruling removed the largest near-term legal overhang on the cabozantinib cash flow and fixed the earliest generic date in early 2030, subject to the November trial. The early-December PDUFA date is the event that separates the current story from the next one, and the stock's doubling off its fall 2025 low reflects a market that has already placed substantial probability on an on-schedule approval.
The judgment is that the franchise is sound, the pipeline has earned clinical credibility, and the capital allocation program is working as intended, but the risk-reward at the current level is no longer asymmetric in the bull direction. The bear case is priced in part, the base case is priced in substantially, and the bull case, a strong STELLAR-304 plus a rapid zanzalintinib ramp plus a 2032 patent boundary, is the only scenario that offers a clear path to a meaningfully higher outcome. The specific variables to watch are the NDA decision, the STELLAR-304 readout, the '039 patent trial, and the quarterly evidence on cabozantinib volume growth in the neuroendocrine tumor indication, where the guidance cut originated. The falsifiable test remains the December 3, 2026 action date, and the quarter that follows it, which should show whether the guidance base can absorb a zanzalintinib launch or whether the buyback program is the only engine left in the per-share story.
The final line of the assessment is that this is a quality franchise with a maturing first drug and a credible second drug, traded at a multiple that has moved ahead of the earnings and that now depends on the next two catalysts landing close to plan. The company has given the market a clear calendar of events, and the discipline of the execution so far, the patent wins, the guidance honesty in August, and the pace of the buyback, supports a view that the base case is attainable. The question is whether the current price pays enough for that attainment to justify the risk that one of the calendar events disappoints. The honest answer is that the margin for a miss is thin, and the margin for a hit is meaningful, which is the definition of a setup that has already had part of its story priced.