BMY at $58.26: A Sum-of-Parts in Three Layers

We value Bristol-Myers Squibb the way the company actually generates value: as a sum of three asset layers, less liabilities and overhead. The first layer is the portfolio of drugs already on the market. The second is the pipeline of drugs in FDA development. The third is the platform — the R&D engine that produces the next round of pipeline assets. Stacking the three and netting against the balance sheet gets you per-share equity value. Each layer is valued on its own merits with its own method.

On pricing. This post publishes intraday on Wednesday, April 29, 2026 — before BMY's Q1 2026 earnings on Thursday morning, April 30. Our reference market price is yesterday's close ($58.26 on April 28). The model-portfolio entry will execute at today's close (April 29) so the position is established pre-earnings rather than reacting to the print. The published thesis sits in front of the market for the trading day before we transact.

The build

  Portfolio of marketed drugs       (Layer 1, 18 products)      $105.4 B
  Pipeline of disclosed Phase 2+    (Layer 2, 18 programs)       +$29.9 B
  Platform value of R&D engine      (Layer 3, ρ × $9.5B R&D/yr)  +$76.1 B
                                                                ─────────
  Operating asset value                                         $211.4 B
                                                                ─────────
  Corporate overhead                (capitalized G&A)             −$6.7 B
  Net debt                          (debt $47.1B − cash $10.2B)  −$36.9 B
  Working-capital and tax adj.                                    +$2.4 B
                                                                ─────────
  Net equity value                                              $170.3 B
                                                                ─────────
  ÷  Diluted shares outstanding                                   2.04 B
  ───────────────────────────────────────────────────────────────────────
  =  Per-share fair value                                          $83

The three layers contribute roughly 50% / 14% / 36% of operating asset value. The breakdown of each is below; we then return to what each layer requires you to believe and what the strip-down to the most defensible parts looks like.

Why SOTP for BMY

A consolidated DCF blends 18 disclosed marketed drugs, 18 Phase 2+ pipeline assets, and a 15-year R&D production cycle into a single revenue-and-margin profile. That hides every disagreement that matters. Eliquis losing exclusivity in 2028 is not the same kind of valuation problem as Cobenfy ramping toward an unknown peak, which is not the same as the question of what BMY's R&D engine produces in 2032. Pharma-SOTP forces these disagreements into the open and lets each one be priced against its own evidence.

Layer 1 — the portfolio of marketed drugs

Eighteen disclosed commercial products. Each is projected through its known LOE year with explicit erosion, valued at WACC = 7.40% (Damodaran-Hamada industry beta of 1.17), 40% EBITDA margin, 18% effective tax. The top six carry roughly half of the layer:

Two facts dominate this layer:

The 2028 cliff is real. Opdivo and Eliquis together are 28% of Layer 1 and both face 2028 LOE. Eliquis is also in the first CMS Medicare-negotiation cohort — IRA pricing is effective January 2026 — and now stacks under the MFN regime as well. The 2028 erosion is modeled with full small-molecule decay; Q1 2026 will give us the first earnings-level read on the IRA + MFN combined hit.

Three ramps are flagged aggressive. Krazati ($0.20B → $2.0B, 58% CAGR), Sotyktu ($0.29B → $4.0B, 55% CAGR), and Cobenfy ($0.15B → $5.0B, 78% CAGR) all require sustained high-CAGR ramps. These peaks track closer to BMS management guidance than to sell-side consensus midpoints. We discuss the subjectivity below.

Layer 2 — the pipeline in FDA development

Eighteen Phase 2 or later assets, valued as risk-adjusted NPV (rNPV) using sector-standard probability-of-success rates by phase and therapeutic area. Top six:

Milvexian had a setback last November (LIBREXIA-ACS arm halted; only AF and SSP arms remain) and the modeled $4B peak already reflects that. The pipeline is structurally healthy in the sense that the largest asset is only 13% of layer 2 — there is no single-blockbuster bet driving this number.

Layer 3 — the platform value of the R&D engine

This layer captures future R&D output beyond the disclosed pipeline. The molecules that haven't yet been announced. The label expansions that haven't yet started. The platform applies a sector-anchored productivity ratio to BMY's recurring R&D budget, with the disclosed pipeline subtracted to avoid double-counting:

Platform PV = (ρ × annual R&D × growth surplus) / (WACC − g) − pipeline rNPV

Inputs: ρ = 1.50x, anchored to the pharma sector mean over the pooled cohort window 2001–2010 R&D → 2013–2022 approvals. Annual R&D = $9.55B, BMY's 5-year mean ex-IPR&D charges. WACC = 7.40%, g = 3.0%.

This produces $76B of platform PV after the pipeline subtraction.

The 1.50x ratio is the median across pharma, not a BMY-specific bullish assumption. It says: if BMY's R&D engine is sector-average, the platform alone is worth $37/share. We deliberately do not adjust ρ per ticker, because we have looked carefully at whether some companies are systematically better R&D operators than others, and the data does not support a per-ticker premium. We documented that exercise — and the 2018→2025 backtest that killed it — in Why Our Pharma Platform Multiple Is the Sector Average. That post also explains why we use a 5-year ex-IPR&D R&D smoothing window rather than a single year (M&A charges are lumpy, recurring research is not).

Where the subjectivity lives

The numbers above are not assumption-free, and we're not going to pretend they are.

Layer 1's subjectivity sits in the four ramping products our own model flagged. Cobenfy at $5B peak is BMS management's implied target out of the Karuna acquisition; sell-side consensus midpoint is closer to $2.5B. The same gap exists on Sotyktu and Krazati. If you swap our peaks for consensus midpoints across those four products, Layer 1 falls from $105B to roughly $90B — about $7/share off the SOTP. We hold the management-guidance peaks because they reflect public commitments under disclosure rules, but a careful reader can knock $7–$10/share off the layer-1 number with defensible alternative assumptions.

The other direction is real too: Eliquis could grow modestly above TTM before LOE (some analysts model peak at $17B+), Camzyos label expansions could push past $3.5B, and Opdivo subQ retention could exceed standard biologic-decay assumptions. Layer 1 could be $5–10/share higher under those.

Layer 2's subjectivity is in the phase-success rates and the peak-revenue assumptions. We use sector-standard PoS tables by phase and therapeutic area. Specific assets could resolve up (admilparant in IPF reads positive in 2H 2026 → +$1.5/share) or down (a milvexian setback or arlo-cel BLA delay → −$1–2/share). The aggregate rNPV is bounded — no single asset moves it more than $2/share — but the layer can drift ±$5/share on the cycle of 2026–2027 readouts.

Layer 3's subjectivity is the ρ ratio itself. Sector-average is the only number we can defend. If you believe BMY is below median (a bear case, given the recent 2018–2025 backtest where BMY ranked dead last), Layer 3 is worth less. If you believe BMY is above median (the bull case from the Karuna integration and CELMoD platform), it's worth more. Each 0.1x of ρ moves the layer roughly $5/share.

We believe the layers are where we have them. They are not point estimates with infinite confidence; they are central tendencies of distributions we can defend with data. The margin of safety conversation — next — is what survives when you stress those distributions.

Sector context

Two healthcare sector forces touch this thesis directly (full sector view in our April 25 healthcare briefing PDF).

MFN drug pricing is now structural. All 17 large drugmakers have signed bilateral US-pricing agreements; BMY is in the basket with Eliquis IRA + MFN stacking exposure. Bloomberg cites a potential ~30% reduction in US drug-company profits under full MFN flow-through. This is a Layer 1 risk: if blended US net price compresses 15% across the in-line portfolio, Layer 1 falls roughly $7/share. Eliquis is doubly exposed.

Patent-cliff M&A and licensing remain the sector's principal capital-allocation theme. BMY is named with bolt-on appetite. Karuna ($14B, 2024) is the most recent example. If BMY can buy Phase 2/3 assets at fair multiples, the platform layer realizes through acquired pipeline rather than fully internal R&D — and the case for accretive bolt-ons stays open.

The April 30 LLY/MRK/AMGN/BMY earnings cluster is the first quantified read on MFN US-net-price flow-through. That's the highest-impact data point on the BMY calendar.

What to watch at Q1 (April 30)

In priority order for the three-layer thesis:

1. MFN flow-through commentary. A >25% blended US net-price hit on signed-MFN products moves Layer 1 down structurally. Watch language on Eliquis given the IRA + MFN stack.
2. R&D guide. Our Layer 3 uses $9.55B annual R&D. A material guide-down rebases Layer 3 (~$8B PV per $1B of R&D); a guide-up with productivity claims supports it.
3. Pipeline progression. Milvexian LIBREXIA-AF timing; admilparant Phase 3 cadence; arlo-cel BLA timing; Cendakimab approval timing. Layer 2 will be marked to whatever moves here.
4. Cobenfy ramp. Q1 prescription and revenue trajectory tests the most aggressive peak assumption in Layer 1.
5. Eliquis Q1 revenue. First quarter under the IRA negotiated price; the question is whether volumes hold.
6. BD posture. What management is saying about deploying balance sheet for further bolt-ons. Direct commentary on whether Layer 3 will realize through acquisition.

We won't re-run the SOTP after the print unless an item materially changes a model input. The point of this list is to score the print against pre-committed checks rather than build a post-hoc narrative.

Portfolio action

We will open BMY at 2% weight in our model portfolio at today's close (Wednesday, April 29, 2026) — funded by selling SPY. Same-day execution is a tighter entry than our usual T+1 convention; the rationale is BMY reports Q1 2026 tomorrow morning, and the post needs to be in front of the trading day before the print rather than reacting to it. The post has been public for several hours before close.

Two percent matches our UNH sizing and reflects a similar conservative posture: the asymmetry is real, but Layer 3 is precisely the layer where modeling uncertainty is highest. Adding aggressively before we have a read on R&D productivity (Q1 print) or label-expansion data (Cobenfy, milvexian) is poor risk management.

Conditional sizing rules: - If Q1 confirms R&D guide and Cobenfy ramp, we'd consider 3%. - If management guides R&D meaningfully lower or pivots away from internal innovation, we'd trim toward 1% or close. - If a flagship Phase 3 asset reads positive (admilparant in IPF most clearly), we'd scale further on layer-2 confirmation.

Current portfolio state in the widget below and at /blog/portfolio.

The margin of safety — strip the business to its parts

This is the close.

Walk down the three-layer build, asking how few of them you need to believe to justify the current price:

If you believe only the marketed portfolio — every drug BMY currently sells, projected to its known LOE, with no credit for anything else — you get $30/share. The business stripped to its absolute floor is worth roughly half the current price.

If you believe the marketed portfolio plus the disclosed pipeline — every drug they sell plus every drug they have in Phase 2+ development, risk-adjusted by sector-standard probabilities — you get $45/share. Still 23% below current price. The disclosed business alone gets you nowhere near the market.

The market at $58.26 is paying $13 above the platform-stripped value. Our model says the platform layer is worth $37/share at sector-mean productivity. The market is implicitly giving roughly 35% credit to the platform. Our model gives full credit. The disagreement isn't whether BMY is cheap; it's how much of Layer 3 you believe in.

The bear case isn't "BMY's drugs disappoint" — that's already in the platform-stripped $45 floor. The bear case is "BMY's R&D engine produces nothing of value over the next decade, the in-line portfolio LOEs as modeled, and the disclosed pipeline plays out at sector-standard rates." If all three hold, the stock should trade around $45. At $58.26 you're risking $13 of platform credit against the bull case of $25 of additional platform credit. That's $25 of upside against $13 of downside before you start stressing Layer 1 and Layer 2 assumptions. The dollar asymmetry is real but modest; what makes the trade interesting is which layer the asymmetry sits in. Layer 3 is the layer the market consistently mis-prices at the sector level (per the companion post on platform productivity), not because anyone has a better view than anyone else, but because there's no defensible per-ticker view at all. You are not betting on BMY being a better operator than peers. You are betting that the platform layer is worth at least sector mean across pharma — including BMY. That bet has a forward-empirical answer rather than a narrative one, and it's what we're sizing into.