The Patent Cliff Is a Slope: What Really Happens to the World’s Biggest Drug in 2028

In December 2023, adalimumab biosimilars held 79 percent of the German market. In February 2024, they held four percent of the American one. Same molecule, same patent status, same science — a twentyfold difference. Anyone trying to price what the largest patent expiry in pharmaceutical history will do to its owner in 2028 has to start by explaining those two numbers.

Because together they show that the industry’s favorite metaphor is misleading. A patent “cliff” implies an edge: a date passes, revenue falls off. That is exactly what happened to Lipitor, to Plavix, to every major chemically synthesized drug of the past three decades. But the wave rolling toward the industry between 2026 and 2030 is overwhelmingly not made of chemical compounds. It is made of biologics. And biologics do not fall. They slide — at speeds that depend on which payment system they are sold into.

This piece does three things. First, it shows that a biologic’s erosion curve is set by a country’s reimbursement plumbing rather than by patent law, which is why the same expiry has a different shape in Germany than in the United States. Second, it shows that two clocks now run on every large American drug — the patent clock and the price clock — and that a single provision in a July 2025 budget law pushed the second clock back by twelve months for two of the world’s biggest cancer drugs. Third, it shows that the revenue lost here does not disappear. It moves: to payers, to biosimilar manufacturers, and — as an acquisition premium — into the accounts of biotech shareholders.

Four numbers for one wave, and none of them is wrong

Analyses of the coming expiry wave carry figures between $170 billion and $400 billion. That is not sloppiness; it is a question of scope — period, geography, and what exactly counts as revenue. All four measures below are defensible. They simply answer different questions.

Framing	Figure	What is actually being counted
Annual sales losing exclusivity through 2032	at least $173.9 billion	The annual revenue of best-selling brands whose exclusivity ends by 2032 — a stock figure, not a cumulative loss.
Branded revenue at risk, 2025–2030	$200–400 billion	A wider net: every product whose protection lapses in the window, with no deduction for what the originator actually retains.
Revenue actually lost	materially lower	Depends entirely on the erosion curve. For biologics, half the peak revenue often survives for years.
The single largest case	$31.7 billion (2025)	Keytruda’s annual revenue. Its core composition-of-matter protection in the U.S. ends in 2028.

The third row is the one that matters and the one that never makes a headline. It contains all the actual analysis: the entire question is what sits between “revenue with an expiring patent” and “revenue lost.” Treat the first number as the second and you will model a manageable, expensive descent as a catastrophe.

The distinction the metaphor swallows

A chemically synthesized drug — a small molecule, a pill — can be copied exactly. The generic is molecularly identical, proving it is cheap, approval runs through an abbreviated pathway, and pharmacists in most states may substitute automatically at the counter. That is why branded revenue collapses within a year.

A biologic is a protein produced in living cells. It cannot be copied identically, only closely — hence “biosimilar.” Development costs hundreds of millions rather than single-digit millions, comparative clinical trials are required, manufacturing is genuinely hard, and automatic substitution requires a separate interchangeability designation that not every product carries. Revenue behaves accordingly.

Characteristic	Small molecule (generic)	Biologic (biosimilar)
Volume lost in year one	80 to 89 percent	30 to 50 percent
First entrant’s discount	80 to 95 percent within 12 months	15 to 30 percent initially
Copier’s development cost	low single-digit millions	hundreds of millions
Competitors after three years	often double digits	usually a handful
Automatic substitution	the norm	the exception, requires extra data
Shape of the revenue curve	a cliff	a slope with steps

One uncomfortable consequence follows for any valuation model: applying generic-erosion assumptions to a biologic is not slightly wrong, it is wrong by years and by billions. But the reverse error is just as costly. Anyone who reads the gentleness of the slope as harmlessness misses the steps — and the steps do not come from the patent office.

Humira, the key witness: four percent share, forty percent revenue loss

The largest biologic expiry to date can be reconstructed in full. AbbVie’s Humira (adalimumab) was the best-selling drug in the world for years, generating $21.24 billion in 2022. U.S. exclusivity ended on January 31, 2023.

What followed contradicts both popular narratives. Revenue did not collapse: global sales fell 32 percent in 2023, with the U.S. down 35 percent, and dropped further to $8.99 billion in 2024. After two full years of competition the originator still held more than 40 percent of peak revenue — unthinkable for a pill.

The revealing part is how that decline happened. In February 2024, after a full year of competition, all Humira biosimilars combined held roughly four percent of the U.S. market. Over the same stretch, in the first quarter of 2024, Humira’s U.S. revenue fell 40 percent. Only one explanation fits both facts: revenue did not fall because patients switched. It fell because the price fell.

That is the essence of the American mechanism. Between manufacturer and patient sit pharmacy benefit managers — intermediaries that run formularies for insurers and derive a substantial share of their economics from manufacturer rebates. An originator that wants to keep its formulary position has to deepen those rebates the moment a biosimilar launches. The list price can hold perfectly steady while the net price the manufacturer actually collects falls off a table. From the outside nothing appears to happen — until the quarterly report lands.

The practical rule for investors: in a biologic expiry, market-share data is a misleading leading indicator. It shows where the patients are, not where the money is. To see the damage early you have to watch net price, and net price does not appear in any prescription database. It appears in the income statement, one quarter late.

79 versus 4 percent: one molecule, two erosion curves

Here is where the international comparison stops being trivia. While U.S. biosimilars held four percent of the adalimumab market in February 2024, they had already reached 79 percent in Germany by December 2023. Same molecule, same period, a twentyfold gap.

The explanation lies entirely in payment plumbing. German statutory health insurers negotiate rebate contracts directly with manufacturers, and physicians are held to biosimilar quotas within regional prescribing-efficiency targets. When adalimumab came off patent, the first biosimilars entered rebate contracts at discounts of up to 40 percent, and prescription steering did the rest within months. Since February 3, 2026, German insurers may also sign exclusive rebate contracts with biosimilar suppliers, sharpening the mechanism further.

This is one of the cleanest natural experiments health economics has to offer. Two markets, one molecule, one window, identical science — and completely different outcomes. The conclusion is unambiguous and central to valuing every affected company: the erosion speed of a biologic is not a property of the molecule. It is a property of the reimbursement system.

Market design	Effect on erosion	Consequence for the originator
Germany: rebate contracts, biosimilar quotas	Very fast volume shift (79 percent inside two years)	Revenue disappears quickly, but the process is short and predictable
U.S.: PBM formularies, rebate economics	Volume stays with the brand; net price falls immediately	Margin erodes before share does — damage surfaces late
Hospital-administered drugs (both)	Faster switching, because purchasing is centralized	Infused products erode earlier than self-injected ones

That last row leads directly to the case at hand. Keytruda is an infused product administered in hospitals and oncology centers — precisely the segment where purchasing is centralized and price advantages transmit fastest. It argues for a steeper curve than the Humira analogy suggests, and it is the single most underweighted factor in the consensus view.

The largest single expiry in industry history

Keytruda (pembrolizumab) is an antibody that blocks a brake signal on immune cells, allowing the immune system to attack tumor cells. It now carries more than forty approved indications and generated roughly $31.7 billion in 2025. In the second quarter of 2026 the Keytruda family posted $8.4 billion, up four percent. Merck guides to $66.3–67.3 billion of total revenue for 2026.

So a single product carries about half of company revenue, and its core composition-of-matter protection in the United States ends in 2028, with European protection running roughly parallel. That degree of concentration is rare and makes the case a template for the whole sector.

One naming caveat worth carrying into any trade: the maker of Keytruda is Merck & Co. of Rahway, New Jersey, which trades in New York as MRK and operates as MSD outside North America — because the Merck name outside the U.S. and Canada belongs to Merck KGaA of Darmstadt, an entirely separate German company. They are not related, and only one of them owns Keytruda.

The market has not ignored the exposure. The stock has traded at a persistent discount to the sector for years, analyst targets cluster around $130 to $137, and the shares trade near that band. The question is not whether the cliff is priced. It is whether its shape is priced.

The second clock: price can fall before patent does

Until a few years ago an American drug had one date that mattered: loss of exclusivity. Since the Inflation Reduction Act it has two. The law lets CMS negotiate prices directly for high-spend Medicare drugs — nine years after approval for small molecules, thirteen for biologics.

The implication is easy to state and widely missed: revenue can fall while the patent is still fully in force. The price clock runs independently of the patent clock, and it can ring first.

That asymmetry has acquired a name — the “pill penalty.” Because small molecules face negotiation four years earlier than biologics, research incentives tilt systematically toward biologics, meaning toward expensive injections and away from pills, even though pills account for more than ninety percent of prescriptions. Industry-aligned analyses put the drop in small-molecule investment since the law took effect at up to 70 percent; that figure comes from advocates and should be read as such, though the direction is not seriously disputed. The bipartisan EPIC Act, which would equalize both periods at thirteen years, has been before Congress since 2025.

Then came the development that drew almost no attention in early 2026 and was worth more to two companies than most trial readouts. The reconciliation law signed in July 2025 widened the orphan-drug carve-out twice over: products designated for multiple rare diseases became ineligible for negotiation, not just single-designation products, and the qualifying clock now starts later for an orphan drug that subsequently wins a non-orphan approval.

Keytruda and Opdivo — both blockbusters with orphan designations behind them — would have qualified for selection in 2026 on spending and statutory criteria. The change pushed their eligibility out by twelve months, to February 1, 2027. Sure enough, when CMS announced the fifteen drugs of the third negotiation cycle on January 27, 2026, with prices effective in 2028, neither was on the list. The Congressional Budget Office puts the cost of the expanded carve-out at $8.8 billion over ten years.

For Merck this buys a year in which the two clocks do not run out together. It also means negotiation and expiry now sit closer to one another, raising the possibility of losing the same product twice in short order — once to a negotiated price, once to competition. Any serious model needs both dates. Most carry only one.

The defense is called Qlex, and it swaps one risk for another

Merck’s answer to 2028 is not a new molecule but a new route of administration. In September 2025 the FDA approved Keytruda Qlex — pembrolizumab combined with an enzyme that makes subcutaneous delivery possible. Instead of a half-hour infusion, an injection takes a few minutes. For patients that is a real improvement; for oncology practices it changes both workflow and reimbursement.

The commercial logic is transparent. The new presentation is protected by its own formulation, device, and delivery patents, potentially running into the mid-2030s. If a large share of patients can be converted before 2028, the expiring composition-of-matter patent lands on a shrunken market: an intravenous biosimilar would then compete not with the market leader but with its predecessor product. Merck targets conversion of 30 to 40 percent of Keytruda volume. Qlex generated roughly $590 million in the first half of 2026 according to reporting on the company’s filings — a credible start, well short of what will matter in 2028.

None of this is novel, and it carries a less flattering name: critics call it product hopping, and the advocacy group I-MAK leveled exactly that charge in 2025. For investors the assessment should be unsentimental, and it cuts both ways.

What conversion buys	What it costs
Several additional years of exclusivity on a slice of revenue, without having to discover anything new.	It replaces a legal risk with a known date with an execution risk with an open outcome: success depends on physicians, hospitals, and payers.
A genuine convenience benefit, which makes the switch clinically defensible rather than purely tactical.	Antitrust and political exposure, in an environment already hostile to drug pricing.
Time — the scarcest input in rebuilding a pipeline.	Payers can push back by preferring the cheaper intravenous form on formulary, which is a one-contract decision.

The modeling point is the one that matters. A patent expiry is a dated event you can discount. A conversion rate is a behavioral assumption about thousands of prescribers. The defense converts a well-forecastable variable into a poorly forecastable one. That can be worth doing — but it widens the distribution of outcomes, and widening the distribution is normally something a market charges for.

The cliff is not a loss. It is a transfer.

Which brings us to the layer that coverage almost entirely omits. Revenue lost by an originator does not evaporate. It is redistributed, to three recipients.

First, to payers and ultimately premium payers. Humira biosimilars produced average savings of roughly $4,505 per patient per year in the U.S. in 2024. This is the intended part of the exercise, and it works.

Second, to the copiers. Their business is the mirror image of the cliff: what one company loses, another collects.

Third — and this is the channel that matters most to shareholders — to the owners of acquisition targets. A company facing the loss of tens of billions in revenue must buy replacements. It has little choice, it has a deadline, and the sellers know both. That shows up in the data.

Metric	Value	What it means
Biopharma M&A value, 2025	$133 billion (+133 percent)	Second-highest in five years, across 50 deals — the most since 2021.
2026 forecast	$140–160 billion	With upside of a further $20–30 billion in a best case.
Acquisition premiums, 2025	60 to 120 percent	80–120 percent for targets with approved products; 40–70 percent for preclinical platforms.
Largest reported transaction	up to $32 billion	Reported advanced talks between Merck and Revolution Medicines — not closed, but indicative of the urgency.

The premiums are the real finding. They are not a mood; they are the price of urgency. A buyer with a patent date behind him negotiates from weakness, and that lands directly in the purchase price. For the acquirer’s shareholder the cliff is therefore a double charge — first the forgone revenue, then the premium paid to replace it. For the shareholder of a plausible target it is an opportunity whose probability can, unusually, be read off a calendar.

How to own the other side

If the transfer is real, the reasonable question is whether it is investable. It is, with caveats that deserve to be stated plainly.

Sandoz, spun out of Novartis, is the world’s largest biosimilar company, with thirteen marketed biosimilars and 28 candidates in development. It is establishing biosimilars as a standalone unit alongside its generics business, explicitly in anticipation of what it calls a “golden decade” of patent losses. It is the closest thing to a direct, diversified way to own the receiving end of the wave, though it trades in Zurich and comes with a 35 percent Swiss withholding tax, of which 20 points are recoverable under most treaties.

Further out on the risk curve sits Formycon, a German developer based near Munich, which has three approved biosimilars (ranibizumab, ustekinumab, aflibercept) and — pointedly — a late-stage pembrolizumab candidate, FYB206, a copy of exactly the molecule that comes free in 2028. Enrollment in its integrated Phase 1/3 trial completed in July 2025, primary endpoint data was expected in the first quarter of 2026, and on February 11, 2026 the company signed an exclusive license with Lotus Pharmaceutical covering commercialization across much of Asia-Pacific in exchange for an upfront payment, milestones, and a share of gross profit.

That structure captures both the model and its ceiling: Formycon develops and manufactures while partners sell. It lowers capital intensity but caps margin, and revenue depends on partners hitting milestones. An investor here is not buying a diversified pharmaceutical company. They are buying a series of binary development and patent-litigation outcomes in a market whose prices fall by design. Position accordingly.

For a U.S. investor there is a further wrinkle worth naming. Biosimilar economics in America are structurally worse than the German comparison implies, precisely because of the rebate machinery described above: a biosimilar entrant must buy formulary access from the same intermediaries the incumbent is paying. That is why several early Humira biosimilars gained approvals and almost no share. The transfer is real, but in the United States a large part of it lands with PBMs and insurers rather than with the copier — which is an argument for owning the payer side, or for owning this theme through a diversified healthcare fund rather than a single name.

Three scenarios through 2030

Rather than a point forecast, the honest output is a range.

Scenario	Path	Early tell
Slope (Humira pattern)	Erosion spread over years; the brand still holds a substantial share of peak revenue in 2030. Subcutaneous conversion partly succeeds and acquisitions fill the gap.	Conversion rate tracks the 30–40 percent target; deals bring approved products rather than early-stage platforms.
Step (hospital-market pattern)	Materially steeper than Humira, because centralized hospital purchasing and several simultaneous biosimilar launches compress the timeline. Price falls before share does.	A growing count of approved pembrolizumab biosimilars; early tenders and rebate contracts in Europe.
Pincer	Price negotiation from 2027–28 coincides with expiry; conversion undershoots; replacements are bought at peak prices.	Selection for negotiation after February 1, 2027; premiums at the top of the range; Qlex share plateauing.

The middle case looks most likely, because it takes oncology’s specific market structure seriously instead of transplanting the Humira pattern wholesale. The key indicator is not the share price but the number of approved pembrolizumab biosimilars: with one competitor the curve is a slope; with five it is a step.

What follows for a portfolio

Four conclusions survive regardless of one’s view on any single stock.

One: revenue concentration is the real metric. Risk is not set by the expiry date but by the share of company revenue a single product carries. A company drawing half its revenue from one drug is a different investment from one drawing fifteen percent, even if both show the same patent year in the table.

Two: ask about the molecule class, not the date. Small molecule means cliff; biologic means slope. Skipping that distinction produces errors in both directions.

Three: carry both clocks. Since the Inflation Reduction Act, the patent date is no longer the first price-relevant event. For two of the world’s largest cancer drugs, February 1, 2027 currently matters more than any patent year.

Four: the other side exists and is investable. The revenue does not vanish, it changes hands. Whether to own that side through a specialist developer, the biosimilar market leader, or not at all is a question of risk tolerance. Whether it exists is not.

The cliff in question turns out not to be a cliff at all, but a slope with steps whose gradient is written not in a patent register but in rebate contracts, formulary decisions, and one paragraph of statute from July 2025. Read it as a date and you will miss it. Read it as a mechanism and you can see it coming — on both sides of the trade.