Damage Math

The near-term earnings hit behind Penumbra's 2025 drawdown was essentially zero — consensus rose and every quarter beat through the 25.8% fall — so the roughly $3.07B of market value it erased had no earnings basis, and the price has since round-tripped in full ($321.19 today, above the $303.76 peak) under a $374 Boston Scientific bid. The one live value question is a modest out-year EPS shave; the temporary-vs-permanent trial split 0.61, contested.

What this tab has to reconcile

The framework's centre asks whether the market treated a small hit to near-term earnings as a large hit to the NPV of all future cash flows. Penumbra is an unusual specimen because that question was posed twice, in opposite directions, and both have largely resolved:

• The 2025 de-rate — a price hit with almost no earnings basis. From a $303.76 peak on 19 Feb 2025 to a $225.54 trough on 5 Nov 2025 the stock fell 25.8% [fit_features:capitulation_gauge.drawdown], erasing ~$3.07B of market cap. Yet through that entire leg consensus rose and every reported quarter beat. That is the classic overreaction setup — and it fully reversed, first by the market (the stock was back above its peak by 31 Dec 2025) and then by a strategic acquirer at $374 [1].
• The 2026 out-year cut — an earnings hit with almost no price effect. The 30 Jul 2026 Q2 print missed on EPS (−14.4%) as SG&A stepped up, and analysts trimmed FY2027/FY2028 EPS while leaving revenue untouched. But the price barely moved: it is anchored 14% below the $374 deal, so the earnings revision has no room to reprice.

Neither episode is a live "small hit, large NPV loss" gap today. The tab below shows the arithmetic behind that conclusion, then hands the residual — is the post-2024 reset a permanent step-down or a timing lull — to the trial, which is where the diagnosis lives.

The numerator: how far did near-term consensus actually fall?

Around the 2025 drawdown, it did not fall — it rose. FY2025 revenue landed at $1,403.7M against a $1,384.4M consensus, and normalized EPS at $3.84 against $3.78; the four quarters reported into and through the fall beat by +8.5%, +24.5%, +4.7% and +6.1%. A near-term estimate that climbs while the price falls 25.8% is the signature of a de-rate, not an impairment — the numerator of any damage calculation over that window is negative.

The only genuine near-term hit came a year later, at the Q2 2026 print (30 Jul 2026), and it landed on margin, not on sales. Q2 EPS of $1.011 missed the $1.182 consensus by 14.4% — the first miss in the series — because SG&A jumped to $391.3M for the half from $313.4M, so first-half operating income fell to $79.3M from $81.2M even as revenue grew 15.3% to $764.8M [2]. Analysts responded by cutting the out-year EPS curve while holding revenue flat:

Source: revision history from data/sp/estimates.json (momentum, 180-day vs current means); revenue in $M, EPS per share.

Revenue held to within a fraction of a percent; normalized EPS was cut 3.5% for FY2027 and 7.2% for FY2028. That is the whole earnings hit — a mid-single-digit shave to the out-year profit curve, top line intact. Note that no near-term guidance offsets it either way: with the deal pending, management withdrew the 2026 outlook and stopped hosting calls [3].

The denominator: the price and EV moves alongside

At 39,291,828 shares [fit_features:market_cap.shares], those prices translate directly into value:

Source: market cap from fit_features.market_cap (shares × close); EV nets ~$427M consensus net cash (data/sp/estimates.json, net_debt FY2025 = −$426.8M). Peak/trough closes from fit_features.capitulation_gauge.

Side by side: consensus near-term earnings did not fall through the 25.8% drawdown, yet the market cap fell $3.07B. The value that vanished at the trough had no arithmetic behind it — and the market agreed within weeks, marking the stock back above its peak.

The NPV arithmetic, two ways

The only earnings damage to capitalize is the out-year EPS shave — FY2028 normalized EPS cut $0.57, from $7.88 to $7.31. I take that as the run-rate hit and run it through two transparent scenarios at a 10% discount rate.

Scenario A — temporary (the cut reverses). Management's own account is that the margin step-down is a completed sales-force build-out, after which "we are positioned to capture sales and operating leverage." If that leverage arrives, the out-year EPS curve re-rates back up and the consensus FCF path — $173M (FY26) → $216M (FY27) → $300M (FY28) — is unimpaired. NPV damage from the near-term hit ≈ $0. Under this reading the entire $3.07B peak-to-trough decline was price damage against zero value damage, and it has already been erased.

Scenario B — permanent (the cut sticks). Treat the $0.57/share FY2028 shave as a durable level shift — $22.4M of annual earning power gone (0.57 × 39.29M shares). Capitalize it:

Workings: annual shave $22.4M ÷ (r − g) for the perpetuities; $0.57 × multiple for the multiple method. Discount rate 10%. FY2028 EPS from data/sp/estimates.json (momentum, 180-day vs current normalized means).

Even the harshest permanent reading destroys ~$0.3–0.5B — roughly 2.5%–3.9% of the current $12.62B cap. Add the already-crystallized Immersive Healthcare wind-down (~$115.3M, $2.93/share, booked in FY2024 and behind the company) and the durable value damage plausibly attributable to the whole episode is on the order of $0.4–0.6B.

The gap

The classic damage-math gap — price overshooting value — is absent at today's price. It existed, and was large, only at the November 2025 trough: a ~$3.07B price hit against a ~$0 near-term earnings hit. That mispricing has been fully arbitraged away, first by the market's own re-rate above the peak and then by Boston Scientific's $374 offer, which values the standalone business 23% above where it peaked before the drawdown [4]. A buyer today is not harvesting a fear-repriced value gap; they are underwriting a 14% merger-arb spread. This is the same read the Dislocation tab reaches from the price side.

The trial: is the post-2024 reset temporary or permanent?

The residual question — whether the 2024 thrombectomy-growth reset and Immersive exit permanently lowered Penumbra's earning-power slope — was tried by two opposing corpus-cited briefs and ruled on by three independent judges. I present each at its strongest; I cannot override the ruling.

The temporary case, at its strongest. The impairment never touched the cash-flow trajectory. At the trough quarter Penumbra raised full-year guidance and grew revenue 17.8% [5]; FY2025 actuals beat across the board; gross margin still expanded, toward a 70% target; the sales-force build that compressed margin is complete; and the forward consensus curve rises — revenue $1,596M → $1,802M → $2,043M and normalized EPS $4.75 → $7.31 out to FY2028. A permanent impairment cannot coexist with a monotonically rising forward curve, and Boston Scientific's $374 bid is external validation that standalone value was intact.

The permanent case, at its strongest. The February 2025 confirmation was that the July 2024 reset was structural, not a timing quarter: FY2024 GAAP operating income fell to $9.3M from $73.6M and 2025 guidance was only 12–14% total growth [6]. The permanence mechanism is lower conversion of revenue to earnings: Penumbra permanently exited Immersive Healthcare; core thrombectomy growth decelerated 20% → 16% → 12.3% (1H26); international mix fell from 28.5% (2023) to 22.2% (2025); and in 1H26 the top line grew 15.3% while operating income fell [7]. The recovery came off a lower base; it did not restore the pre-reset slope, and the out-year EPS cuts are the proof.

The ruling. The judges put the probability the impairment is temporary at 0.61, with a mean of 0.57 and a wide 0.26 spread — the panel is contested. The split is reading-order sensitive: the two judges who read the permanent brief first landed at 0.61 and 0.68, while the judge who read temporary first landed at 0.42, a 0.225 order gap. I present both readings and pick no winner. The diagnosis probability this report carries is the trial's 0.61, not my own.

What would break the tie, per the panel: FY2026 operating income growing faster than revenue (leverage finally arriving after the "build-out complete" claim), thrombectomy re-accelerating back above ~16%, and FY2027/FY2028 revisions turning back up rather than continuing to slip — each would tilt temporary. The mirror conditions — SG&A staying above 50% of revenue, thrombectomy stuck near 12%, international mix stuck at 22% — would confirm permanence.

Which line broke, and whether it self-corrects

The break is narrow and specific: thrombectomy volume growth and SG&A leverage, not price or the top line. Core thrombectomy decelerated from ~20% (FY2024) to 16.2% (FY2025) to 12.3% in the first half of 2026, while embolization and access held up at 21.8% [8]. Pricing was "substantially unchanged," so growth now depends on harder volume conversion, and the embolization sales-force build pushed SG&A to 51% of revenue before the promised leverage appeared. FY2025 still produced $189.2M of operating income on 17.5% revenue growth and a 67.1% gross margin [9] — the engine works; the question is the slope from here.

The self-correction mechanism, if it exists, is operating leverage on a completed cost base plus new-market catalysts (Thunderbolt neuro, STORM-PE, CAVT). The structural case against is that the growth algorithm has permanently reset lower — a decelerating core, a durably U.S.-heavy mix after the international and China resets, and a commercial cost-to-grow that keeps SG&A elevated. The corpus does not resolve which; the trial's 0.61 is as far as the evidence carries, and the $374 deal — should it close at price — would retire the question before FY2026 results can settle it.