A non-consensus thesis: the forthcoming demise of Bristol Myers Squibb.

A first proof point.

This is the first of what may be many blog posts about Bristol Myers Squibb. My thesis — outlined below — is that

1. Bristol Myer faces enormous loss of revenue from patent cliffs

2. Bristol Myers bought back too many shares at high prices (over the years 2019-23 BMY repurchased $28.3 billion in stock)

3. Bristol Myers has completely stopped stock repurchases primarily because they have a lot of debt and need the money (and then some) to fund the acquisitions of new drugs to fill their revenue hole; and

4. They have botched their very large capital expenditures so badly that they will not come close to filling their revenue hole.

This is a strongly-held view — and it may be wrong. But state it up front: The problem at Bristol Myers is so acute that Bristol Myers will likely lose their independence as a major company. At some point Bristol will exist to service their debt load. Shareholders will get little.

In the worst case (and worse than our bearish models predict) Bristol Myers will file bankruptcy and be restructured.

The first three parts of this thesis are uncontroversial. The patent cliff is well known. The buy-backs, the debt, the stopping of buy-backs are all visible to anyone who looks at the accounts.

It is the fourth part — that they botched their capital expenditures very badly — is the controversial thesis. And there are good reasons to think I am wrong.

Bristol have teams of PhDs and massive experience in marketing new drugs. Over the last century they have done it well.

By contrast I run a small fund based above a shopping mall in Sydney. I have experience in biotech investing and in going long and short drug companies but I do not have the expertise that a large pharmaceutical company has. And I have been wrong plenty of times about drug companies.

The reason for this post is that we have a fast-approaching partial proof point on the controversial part of our thesis. There will be presentations at the (hugely important) American Society of Clinical Oncology (ASCO) conference that we think will show that many billions of dollars of Bristol Myers money was wasted.

It should be noted that even this is only a partial proof point. This one event is not a solvency event for Bristol Myers. But it will, we think, be part of a pattern that emerges where Bristol Myers took the money that was necessary to ensure their survival and squandered it.

That is why I think there will be many blog posts.

An objective observer will probably think that I am out over my skis. After all, Bristol is a drug company and I am closer to a suburban accountant. What should I know?

But in my arrogance (which may be unwarranted) I think I know a bit. And this post will be the first demonstration.

Part a: Bristol Myer Squibb’s patent cliffs

At the risk of stating the obvious, Bristol Myers has a patent cliff problem.

As any pharmaceutical analyst knows, patents are time-limited; and once they lapse, competitors are free to produce identical or highly similar copies of the original molecule and market them at a fraction of the cost.

With a blockbuster small-molecule drug, patent expiration typically results in near-total loss of the franchise. Small molecules are usually easy to manufacture and distribute, and competitors come. Pfizer — once the biggest drug company — have never really recovered from the expiries of the patents on Lipitor and a few other blockbusters.

Large-molecule drugs tend to survive patent expiry better, for many reasons including manufacturing cost, and proving full substitutability, and other reasons. Typically, when a large-molecule patent expires, the drug company that owns it will keep 30-50% of the revenue — though even this tends to decay over time.

Large pharma companies operate on a treadmill and must continuously discover (or more often buy in) new blockbusters just to backfill the erosion from products going off-patent. The US pharmaceutical industry is currently facing a patent cliff of ~$180 billion (between now and 2030) on industry revenue of ~$500 billion in FY2025.

Bristol Myers is famously exposed. Multiple sell-side analysts have identified them as having potentially the largest exposure to “loss of exclusivity” (LOE) relative to their revenue, with more than 50% of their sales between now and 2030 at risk.

Their exposures include but are not limited to:

- A royalty on Keytruda, $588 million in FY2025, which will disappear completely on 31 December 2026 per their agreement with Merck. (This royalty has a 100% margin so is an important part of profitability.)

- Revlimid, a small molecule with $2.9 billion of sales in FY2025. Unrestricted competition came during the first quarter of this year and sales fell precipitously. These sales will go away almost entirely.

- Eliquis — which is a blood thinner with a very large market mostly amongst elderly people. This is currently the only substantial driver of sales growth. Alas, this giant revenue line ($14.4 billion of sales in FY2025) will go away almost completely. Generics are anticipated in the US from April 2028. Effective 1 January 2026, the negotiated price of Eliquis in the US Medicare Channel has been reduced by ~56% (to $231 per 30-day supply from a $521 list price) per an agreement with the Centers for Medicare & Medicaid Services (CMS) under the Inflation Reduction Act’s Medicare Drug Price Negotiation Program.

- Opdivo, $10.0 billion of sales in FY2025. Bristol Myers is expected to lose exclusivity in the US in 2028 and in Europe in 2030. This is a large molecule and Bristol can expect to keep some of this revenue.

- Pomalyst, $2.7 billion of sales in FY2025. Generics have already entered the market in the US and EU. Effective January 1, 2027, the negotiated price of Pomalyst in the US Medicare channel will be reduced by 60% — to $8,650 per 30-day supply from a 2024 list price of $21,744 — per another agreement with CMS under the IRA Medicare Drug Price Negotiation Program (second cycle / IPAY 2027).

Part b: Bristol Myers’ balance sheet

I am linking to Bristol Myers’ accounts for the past 20 years both as presented and as standardised by the ever-useful S&P Capital IQ.

You can see the buy-backs ended very sharply, and despite lower share prices the company did not continue buying back shares.

But they did acquisitions, including almost $22 billion in 2024 and almost $4 billion in 2025. There were also funding commitments for R&D at third parties. It is one of those commitments — to BioNtech — which is the subject of this note.

Debt now stands at $46.4 billion and net debt at $35.5 billion.

This debt does not look hard to service. Operating cash flow for the last twelve months was $13.3 billion. That is falling, and absent new sources of revenue it will fall catastrophically. But provided the acquisitions deliver, servicing this debt should not be a problem.

Part c: Bristol Myers had to stop buy-backs to fund business development

I assert that Bristol stopped buy-backs to focus on bringing new drugs to market. This is uncontroversial and consistent with the statements of the company.

In commentary at the Bernstein Conference in May 2025, CEO Christopher Boerner stated that while the company retains the ability to execute buy-backs, doing so is a “lower priority at the moment” compared to business development and maintaining the dividend.

There have been many acquisitions — some small, some large — of which Karuna was the one that focused my attention. Karuna was a large-cap startup making a schizophrenia drug that has now launched. We were short Karuna, thinking the drug had many marketing problems (maybe the subject of a future blog post). This was a mistake: Bristol Myers bought Karuna for $14 billion (about $12.7 billion net of acquired cash).

Since 2023, Bristol Myers’ net debt is up from $28.9 billion to $35.6 billion despite stopping buy-backs. Dividends (roughly $5 billion per year) and acquisitions have driven debt up.

Part d: How has Bristol Myers handled their business development?

The three parts above are uncontroversial. Bristol Myers has a big revenue problem. They have a large and increasing amount of debt. They have stopped buy-backs so they can fund “business development” (which really means the purchase of new drugs to fill their revenue hole).

The controversial part of my thesis is that this business development has been botched on a large scale.

Certainly the company has not stated this. Rather the opposite:

Despite flat revenue for the year and these significant loss-of-exclusivity exposures, at the JPM conference in January 2026 the CEO of Bristol Myers described the company as being “built for growth” and expressed “very high” confidence in the company’s ability to overcome these headwinds.

So the big question for Bristol Myers is what do they have in the pipeline that can replace this revenue?

The short answer is: a lot of things (which is why we think this is a series of blog posts). They are spending a lot of money — possibly enough to bankrupt them if it does not work — but with that money they get multiple shots on goal.

We think these big bets will mostly miss, but it will take time to prove that out. If these shots mostly miss then Bristol Myers will lose its independence.

The BioNtech bispecific

This week, however, we will get a sneak-preview of results that are directly relevant for one of Bristol’s most ambitious programs.

Pumitamig (BNT327/PM8002), is a PD-L1/VEGF bispecific licensed from BioNTech for $3.5 billion in non-contingent payments and up to $7.6 billion in additional development, regulatory, and commercial milestones. (This is to own 50% of any eventual drug.)

I think this money will be torched, and I should explain why.

In an ideal world, cancer cells — which are different to normal cells in lots of ways — should be recognised and destroyed by the human immune system. The problem is that many cancers develop mechanisms of avoiding detection or turning down attacks from the immune system. One of the most common mechanisms is the so-called “PD-1/PD-L1 pathway”.

Many tumour cells express a protein on their surface known as “PD-L1”. When that protein comes into contact with the “PD-1” receptor on the surface of T cells, it suppresses the immune system’s natural ability to kill tumour cells. Many people talk about this pathway as being a “brake” on the immune system.

It has been demonstrated that blocking the interaction between PD-1 and PD-L1 with a drug (by inhibiting either PD-1 or PD-L1) lifts this “brake” off the immune system (particularly T cells) — and increases the probability of a tumour response (shrinkage of the cancer) and an increase in survival rates for patients across a wide range of cancer settings.

There are now around 15 approved monoclonal antibodies that work in this manner and together generated approximately $55 billion of sales in 2025. By far the biggest of these is Merck’s Keytruda, which generated $32 billion of sales in 2025. Keytruda was the biggest annual-revenue drug in history until Mounjaro/Zepbound (the weight loss drug from Lilly) knocked it off.

VEGF inhibitors are another major class of cancer therapy. They work by controlling the rate at which your body produces capillaries. The idea is that you zap a cancer using radiotherapy or similar and that wounds the cancer. A VEGF drug will stop capillaries forming thus limiting the ability of the cancer to feed itself and grow back. These were amongst the biggest drugs in the world 15 years ago. The best-known drug is Avastin.

Bristol Myers’ thesis is that PD-L1/VEGF bispecifics will largely “replace PD-1, PD-L1 inhibitors” and will “transform the current landscape as a backbone of cancer treatment”.

They believe pumitamig, their PD-L1/VEGF bispecific, has “multi-billion-dollar potential” and is “poised to transform the standard of care” in over 10 tumour types. They have initiated an expansive development program, including 8 registrational studies by the end of the year, and expect to launch the product in seven conditions by 2030.

They are so bullish they are already scaling up the development of this drug, per the Q4 2025 conference call:

Now when you look, we are actually already scaled up the development of this drug. The confidence is very high. We believe -- the strategy is very simple. We want to replace and then we want to expand. We want to replace where PD-1, PD-L1 inhibitors are playing today in those indications through this bispecific. And then we want to expand because we believe bringing VEGF on top of PD-L1 inhibition we can also tackle some of the indication where PD-1, PD-L1 inhibitors are not working well enough or not at all.

To achieve this goal, Bristol need to prove that combining these two mechanisms into one molecule is better than:

(a) a PD-1/PD-L1 inhibitor alone, which is a very common regimen in oncology and often the standard of care (with or without chemo).

(b) administering a PD-1 or PD-L1 inhibitor and a VEGF inhibitor separately. This combination is routinely used in some settings but is in theory available to any oncologist that deems it appropriate for their patient. (Note: with Keytruda and Avastin out of patent this separate route would be much cheaper.)

For the first question: it is worth noting the chequered history of anti-VEGF agents in oncology. While some anti-VEGF agents have improved overall survival in some settings, this has largely not been the case. By and large (in at least 50 trials across a multitude of anti-VEGF agents), these agents have produced improvements in “progression-free survival” without any improvement in overall survival. It is not known whether this mismatch is explained by a real reduction/slowing in the tumour followed by a rebound that cancels any prior benefit, or whether the VEGF agents are merely producing a radiographic illusion due to changes in the vasculature of the tumour.

PD-(L)1 combined with VEGF has also been tested by various people and has produced disappointing results. There is an argument that combining the two mechanisms into a single molecule will improve “cooperative binding” (specificity), but it could also lead to dosing mismatches that lower efficacy or increase toxicity, and it is not clear that VEGF and PD-1 are co-located in the body in a way that actually leads to better targeting of the tumour cells.

So far Bristol have not announced plans to conduct a head-to-head study answering the second question.

Akeso is running Harmoni-6, a Chinese-only trial in non-small cell lung cancer, comparing a combination of their PD-L1/VEGF bispecific (ivonescimab) with chemotherapy to a combination of BeiGene’s PD-1 inhibitor with chemotherapy. It has already been announced that there is an improvement in progression-free survival. Akeso have been awarded one of five slots in the plenary session at ASCO later this month to announce the overall survival results of the trial.

At ASCO we should find out whether PD-L1/VEGF bispecifics represent a real paradigm shift or a failed spin on an old idea.

Bristol management have expressed that their conviction in pumitamig comes partly from the data that has been generated from other drugs in the class, including Akeso’s ivonescimab, and that Akeso’s upcoming ASCO presentation will serve as useful external validation. Bristol have also expressed a belief that unlike the history of anti-VEGF agents in oncology, the progression-free survival benefit generated by these bispecific agents will translate into longer overall survival.

If Akeso produces a positive benefit on overall survival, then great: Bristol have external validation, though they will still face a lot of competition. There are more than 35 PD-(L)1/VEGF bispecific antibodies currently in development, including more than a dozen product candidates in clinical development.

But if Bristol’s thesis is wrong and Akeso does not produce an overall survival benefit, will they conclude that their thesis is broken, terminate all further development of pumitamig and write off the purchase price? Somehow I doubt it. I do not think this company is much good at big capital allocation decisions.

But investors might prudently write it off. And might question whether Bristol and its current CEO are the right people to spend the tens of billions of dollars on business development that Bristol is spending.

We don’t doubt that Bristol needs to spend money. The revenue hole they have is enormous — and unless they generate lots more revenue they will fade into obscurity. But pumitamig is a drug on which Bristol has spent billions of dollars and on which it has pinned much of its future growth.

And we are about to find out whether this is real or whether it is bullshit. Akeso will tell us. Our guess is that Akeso produce an improvement in progression-free survival (as already shown) but no overall survival benefit. [Note an improvement in progression-free survival without an improvement in overall survival is clinically meaningless.] Further, this progression-free survival benefit is a result of VEGF inhibitors shrinking the (apparent) size of a tumour without actually impairing the tumour.

But what should be really embarrassing for Bristol is that a few guys above a shopping centre on the arse-end of the world should work it out as they pissed away billions.

If I am right there will be more proof points and more posts. If I am wrong, well, I am probably just over my skis.

John

PS. Disclosures are warranted. I am short Akeso. And I am short a lot of Bristol Myers.