BRISTOL MYERS SQUIBB CO (BMY) stock intrinsic value fair value estimate & owner earnings analysis

Bristol-Myers Squibb sells prescription medicines. In 2025 it booked $48.2 billion of revenue, about 69% of it in the United States. Two drugs carry the business today. Eliquis, the blood thinner it sells with Pfizer and whose profits it shares, brought in $14.4 billion. Opdivo, the cancer immunotherapy, brought in $10.3 billion including its new under-the-skin version, Opdivo Qvantig. Behind them sit Orencia ($3.7B) and Yervoy ($2.9B), which are mature. The younger medicines are still growing: Reblozyl ($2.3B, +31%), Breyanzi ($1.4B, +82%), Opdualag ($1.2B), Camzyos ($1.1B, +77%), Sotyktu, Krazati and Cobenfy. Together these seven brought in $6.6 billion. Older drugs are fading fast. Revlimid and Pomalyst brought in $5.7 billion together in 2025, down from $9.3 billion in 2024. The dates that matter come from the company's own patent table. Under settlements with generic makers, U.S. generic Eliquis can launch in 2028. In Europe, Eliquis patent protection ends in November 2026, and some generics are already on sale there. Opdivo's key U.S. patent runs out in 2028, in Europe in 2030 and in Japan in 2031. Opdivo Qvantig's patent estimates are the same, because pending Qvantig patents are not counted. From the start of 2026, Revlimid generics no longer face volume limits, and U.S. Pomalyst generics were expected in the first quarter of 2026. Most of the younger drugs are protected until 2031-2037. How we measure earnings. We start from cash generated by operations. We remove working-capital timing swings. We subtract stock pay as a real cost. We subtract the capital spending needed to keep the business running. Large non-cash charges are added back, such as amortisation of acquired drug rights and the write-off of about $13 billion of acquired research in 2024. They are sunk costs. The decline of these patent-protected assets is instead charged in the forecast, product by product, as each patent expires. We value the business before its lenders are paid, so we add back cash interest ($2.1B in 2025), and then subtract the debt separately. On that basis cash earning power was very steady from 2021 to 2025, between $7.05 and $7.80 per share, with a median of $7.28. That is about $14.9 billion a year, and it is our starting level. We begin in 2021 because that is the first full year after the Celgene deal reshaped the company. Per-share owner earnings after interest were $6.25 in 2025, about flat from $6.27 in 2024. Before interest, per-share cash earnings grew 1.8% in 2025. Our base case assumes about 2.7% for 2026, a little higher, because the company's report for the first half of 2026 (filed July 30, 2026) showed revenue up 4.2% to $24.5 billion. Only the last quarter of 2026 is still ahead in our valuation. First-half 2026 operating cash flow fell even though profits rose. Most of that came from about $3.1 billion of accrued liabilities being paid down (largely rebate timing) and a $0.5 billion rise in receivables. That is timing, not lost earning power. What we credit and what we leave out. We give credit for the younger drugs' growth only as momentum they are already showing, and we taper it to zero by 2030. We then let each drug fall away as its protection ends. Before 2029 we give no credit to anything still in the pipeline. The company spends about 22% of revenue on research, and we charge that cost every year, scaled to revenue. Its long record of turning research into new medicines (Orencia, Sprycel, Yervoy, Eliquis, Opdivo, Opdualag, Sotyktu) shows the spending does produce replacements. So from 2029 we credit new-product revenue of $1.0 billion a year in the base case, $0.4 billion in the bear case and $1.5 billion in the bull case. That is at or below its demonstrated pace. We do not credit management's $2 billion cost-savings program in the base case, because the company says the savings will be reinvested. We do not credit any pipeline drug by name, a recovery in the share price, or further acquisitions. Base case (worth about $52.73 per share). Eliquis revenue drops to about $12.2 billion in 2027 as European generics spread. It then falls roughly 80% after the 2028 U.S. generic launch, following the path Revlimid took once its generics were unrestricted. Opdivo holds near $11 billion through 2027. It then erodes more gradually, as biologics usually do, to about $3.8 billion by 2033. Orencia, Yervoy and the other mature drugs shrink about 3% a year. The younger drugs grow about 30% in 2026 and 10% in 2027, level off by 2030, then decline as their patents expire from 2033. Total revenue falls from about $49 billion in 2026 to about $25 billion in 2035. Yearly cash before interest falls from about $15 billion to about $8.6 billion. After 2035 we let that level shrink 2% a year: the research engine replaces most, but not all, of what expires. The business is worth about $138.2 billion ($67.49 per share). It owes a net $30.2 billion ($14.76 per share). The shareholders' stake is therefore about $108.0 billion, or $52.73 per share. Bear case (about $18.70 per share). European Eliquis generics spread faster, and U.S. erosion after 2028 is steeper. Opdivo biosimilars take share faster. The mature drugs shrink 7% a year, and the younger drugs stall after 2027. Research replaces only $0.4 billion of revenue a year. The case also charges $1.5 billion a year of deal spending that earns nothing, reflecting a documented pattern: about $21.8 billion spent on acquisitions in 2024 alone, much of it written off as acquired research within the year. After 2035 the remaining cash shrinks 6% a year. The business is worth about $68.5 billion ($33.46 per share), less $30.2 billion owed, which leaves $38.3 billion, or $18.70 per share. Even here, yearly cash before interest stays at least about 2.2 times the interest bill through 2035. Bull case (about $94.42 per share). Eliquis and Opdivo erode at the gentler end of past experience. The younger drugs keep more of their momentum, growing about 34% in 2026 and tapering to flat by 2031. The cost program adds about $0.8 billion a year from 2027. Research replaces $1.5 billion of revenue a year, which holds cash earning power at around $13 billion a year after the 2028 drop. The business is worth about $223.6 billion ($109.17 per share), less $30.2 billion owed, which leaves $193.4 billion, or $94.42 per share. The debt. At June 30, 2026 the company owed $43.1 billion of borrowings. We subtract that and add back about $8.5 billion of cash and securities beyond an assumed $3 billion needed to run the business. Cash and securities totalled about $11.5 billion; the mid-2026 securities balance is estimated from the year-end figure plus first-half purchases. We also subtract the $2.0 billion of anniversary payments owed to BioNTech in 2026-2028 under the June 2025 partnership, worth about $1.8 billion today. Most of the debt is long-term fixed-rate bonds: only about $8.9 billion matures through 2030, and some bonds run past 2060. The average coupon is about 4.1%, well below the 6% we discount at. Borrowing that cheaply for about 15 more years on average is worth about $6.2 billion to shareholders, which lowers the effective debt to $30.2 billion. The company is rated A2/A and has no financial covenants on its $5 billion credit line, extended to January 2031. We see no sign that refinancing will be a problem. What changed since our last review. Our previous estimates were $63.82, $110.83 and $151.39 per share. They valued earnings after interest and never subtracted the roughly $43 billion of debt. The debt was mentioned only as a weakness in the bear case. We now value the business first and then subtract what it owes, which by itself takes about $14.76 off every case. The previous base case also had earnings recovering after 2028 as younger drugs took over. We no longer credit that recovery, because it depends on drugs and markets that are not yet in hand. Some things moved up. Before-interest earning power is higher than the after-interest figure used before. First-half 2026 revenue came in stronger than we had assumed. And the low fixed coupons on the bonds reduce the effective debt. What would make this wrong. On the downside, an earlier Eliquis generic launch or a court loss in Europe, or faster uptake of Opdivo biosimilars. Another large acquisition that does not pay off would also hurt. On the upside, a major pipeline success such as a new blockbuster from the research engine could lift value well above the base case. So could Opdivo Qvantig winning patent protection beyond 2028.