The textbook version of generic price erosion says the price war ends fast: a molecule loses patent protection, a wave of ANDA approvals hits within a year, and the price collapses 80–90% before settling into a permanent floor. What CMS's own acquisition-cost data shows is messier — and more useful. Pulling weekly National Average Drug Acquisition Cost (NADAC) records for five oral generics spanning 20 years of post-exclusivity age, we found that the erosion curve does not track years-since-cliff on any fixed schedule. Sildenafil (generic since 2017, ~9 years post-cliff) and aripiprazole (generic since 2015, ~11 years post-cliff) lost 25–35% of their acquisition price in just the six months from December 2025 to June 2026 — an annualized pace well over 50%. Atorvastatin (generic since 2011) and simvastatin (generic since 2006), both well over a decade into genericization, had nearly flattened, moving less than 4% over the same window. Tadalafil (generic since 2018, ~8 years post-cliff) — the youngest generic in the set — was flatter still, ticking up about 1% rather than falling at all.
What the numbers actually say
NADAC (CMS's survey-based acquisition-cost benchmark) publishes a per-NDC average price roughly monthly, drawn from a voluntary retail-pharmacy survey. Comparing the published per-unit price for a single strength/form of five widely dispensed generics, at the reporting window closest to December 2025 vs. June 2026:
Why the tail is so long — and why it isn't just about age
The mechanism is manufacturer entry, not time. FDA's own Generic Competition and Drug Prices analysis found the first generic entrant cuts price roughly 31–39% versus brand (depending on whether AMP or invoice prices are used), two entrants reach about 54%, and it takes six or more ANDA holders to push erosion past 95% — the deep discount typically associated with the "cliff." Getting from one manufacturer to six-plus is not automatic or fast: it depends on how many firms find the molecule's manufacturing complexity and margin attractive enough to file, and low-volume or technically fiddly products can sit at two or three suppliers for years. Aripiprazole and sildenafil are both large-volume, high-visibility molecules that still appear to be attracting new ANDA entrants years after their first launch; simvastatin and atorvastatin, by contrast, have had 15–20 years for every economically rational manufacturer to already be in the market, so incremental entry — and incremental price pressure — has largely exhausted itself. Tadalafil is the outlier that actually reinforces the mechanism: despite being the youngest generic in this set, its 6-month price was flat, which is easier to explain by manufacturer count already having saturated during its launch wave than by its age — the same dynamic that eventually stalls out every molecule's price war, just on its own competitor-driven timeline rather than a fixed calendar.
Caveats
NADAC is a survey average, not a transaction census — it excludes mail-order and specialty pharmacy, and rebates/off-invoice discounts aren't captured at all, so absolute levels understate what large chains actually pay net of rebate. Our comparison window is only six months (Dec 2025–Jun 2026) inside a live index that itself only carries roughly 18 months of weekly snapshots, so we're reading the current slope of five curves, not re-deriving the full multi-year trajectory from each drug's actual patent-cliff date. And single-strength, single-form comparisons (e.g., one tablet or solution strength only) can diverge from a molecule's overall market if generic entry is uneven across strengths.
Why this matters: for payers, PBMs, and generic manufacturers modeling margin, "erosion stops after year one or two" is the wrong assumption for high-volume oral solids — real acquisition-cost data shows meaningful price pressure can persist 9–11 years past first generic entry for some molecules, while a molecule half that age can already be flat. Competitor count, not years-since-cliff, is the variable to track. Anyone pricing a new ANDA launch, or forecasting Medicaid reimbursement floors, should be watching manufacturer entry on the specific molecule, not assuming a uniform age-based curve.