Schrödinger (SDGR) Zen Horizon Research Report

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow large is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?4/10

  A

  Conclusion first: on this dimension, Schrödinger is in a “large market runway, but medium and already slowing reachable ceiling for itself” position. It is expanding an existing pie, selling R&D tools to pharma companies, developing drugs itself, and taking equity economics, all of which are existing markets. It is not creating an entirely new market. What Baillie Gifford’s question is really testing is “the ceiling that SDGR as a company can effectively reach,” not “how large computational / AI drug discovery is as a category.” Once those two are separated, the conclusion is clear: the category is a genuine long runway, but SDGR’s own share of it is, in name, only around a low-single-digit percentage of pharma R&D spending, and its ballast leg, software, is already sending a signal about its “own ceiling” with ACV of about 200 million, growth down to 10–15%, and FY26 consensus revenue at -6.5%.

  1. Nominal category ceiling (TAM): high, but definitions are messy and easy to overstate. “Computational / AI drug discovery” is indeed a long runway. Third-party market studies size the AI drug discovery market in 2026 anywhere from about USD 2.9 billion, 3.6 billion, to a higher-end 8.6 billion, while long-term forecasts to 2035 range from 4.4 billion up to USD 16.0 billion, with CAGR of 14%–23%. The submarkets closer to SDGR’s business essence, molecular modeling / computer-aided drug discovery (CADD), have more convergent definitions: the molecular modeling market was about USD 7.24 billion in 2024 and about 19.5 billion in 2032 (CAGR ~13%), while in-silico drug discovery was about 3.4 billion in 2024 and about 12.8 billion in 2034. The underlying support is global pharma R&D spending already above USD 300 billion/year. On those numbers alone, the category is plainly a “real long runway.” The trap is that these TAMs include CROs, wet labs, AI black-box screening, automation, data, and many other links that SDGR does not directly cover with software sales.

  2. SDGR’s own serviceable market (SAM/SOM): far narrower than the category, nominally only a low-single-digit percentage of pharma R&D spending. The key “demystifying” data come from estimates focused on SDGR itself: if the computer-aided drug discovery market is treated as about USD 7.5 billion today and about USD 12 billion within ten years, then the TAM for Schrödinger’s software platform is roughly only around 3% of global pharma R&D spending by 2029, and that is still an optimistic framing that gives it the entire CADD software pool. In reality, SDGR’s software leg has annual ACV of only about USD 198.5 million, and FY2026 ACV guidance is only USD 218–228 million (+10–15%). In other words, even under the ideal assumption that “the full CADD software pool belongs to us,” today’s penetration is still only single digits. A high category ceiling does not mean SDGR itself gets that ceiling. That is exactly what Baillie Gifford’s question is meant to puncture: serviceable market (SAM) and category TAM are two different things.

  3. Is it “expanding an existing pie” or “creating a new market”? The answer is clearly the former. Lay out the three legs, and none is opening a market that did not previously exist:

  Three legs	Who it sells to / how it earns	New market or existing market
  ① Software licensing (FEP+ / Maestro / LiveDesign)	Sells computational drug discovery tools to pharma companies (19 of the top 20)	Existing market, competing with Certara, Simulations Plus, and OpenEye/Cadence for the same “pharma R&D tools” budget
  ② Drug discovery (owned + partnered pipeline)	Develops drugs itself, collects milestones and royalties	Existing market, competing with all biotech / large pharma in the same new-drug R&D market
  ③ Equity portfolio (Nimbus / Morphic / Ajax)	Receives proceeds when co-created companies are acquired	Existing market, essentially monetization of early-stage biotech equity investment / incubation

  Its differentiation is “using physics, the better shovel represented by the FEP+ gold standard, to mine an existing gold mine,” not inventing an entirely new gold mine. Even if FEP+ is treated by the industry as the “gold standard for predicting binding affinity,” it reaches an already existing pharma R&D software market. It is expanding its share of an existing pie, not creating incremental market demand. The one attempt that could truly “create a new market” is pushing the agentic AI product Bunsen (early access in summer 2026) down to a broader user base and raising platform utilization. But that is still “expanding the reachable boundary of an existing software market,” not opening a new category, and it has not yet been proven.

  4. Horizontal comparison: the ceilings of similar “pick-and-shovel” companies confirm that SDGR’s pie is not especially large. Players in adjacent pharma software / biosimulation markets report TAMs in the same order of magnitude as SDGR, or not smaller: Certara reports TAM of about USD 12 billion; Simulations Plus has a core addressable market of about USD 700–800 million (roughly doubled to about USD 8 billion after acquisitions); the overall biosimulation market was about USD 2.8 billion in 2022 and about USD 9.8 billion in 2030. In other words, the entire “computational / simulation software sold to pharma companies” pie is only a low-teens-billions scale in aggregate, and it is fragmented with many players. This means that even if SDGR’s software leg becomes an industry leader, its nominal ceiling is still “a slice of a several-billion-to-low-teens-billion niche software market,” not the kind of ultra-long runway Baillie Gifford prefers, one that can support nonlinear, order-of-magnitude expansion over the next decade.

  5. Honest landing point: a high category TAM does not equal a high ceiling for SDGR itself, and the slowdown in the software ballast leg is itself the signal of its “own ceiling.” This has to be stated plainly: a high category TAM is a relatively hollow bullish argument. What Baillie Gifford really wants to see, “the ceiling SDGR can effectively reach,” is capped by three facts. First, the most certain leg, software, has seen ACV growth slide from 20%+ in earlier years to 10–15% in guidance, which is precisely evidence that in the core market where it should be “expanding the pie,” expansion is converging rather than accelerating. Second, FY2026 consensus total revenue is about -6.5% negative growth (because of a tough base, but negative growth itself weakens the “ceiling is far from reached” narrative). Third, drug discovery and equity monetization have upside elasticity, but both are lumpy, unschedulable pulses that cannot provide a compounding engine that “keeps pushing the ceiling higher.” The category is a long runway, but SDGR’s effective climbing speed on that runway is slowing. A ballast leg whose growth has fallen to the mid-teens and whose revenue is expected to decline in the near term is itself a sign that “its own ceiling has been seen by the market,” not that “the market has failed to recognize its greatness.” This is consistent with the report’s judgment of “solid downside floor, upside awaiting catalysts, Hold rating”: it is a good business with real share and a real moat, but measured by Baillie Gifford’s yardstick for “a great growth stock that can rise 5 times over the next ten years,” its reachable ceiling and growth slope deserve only a medium score.

  Jun 5, 2026
• QCan its revenue at least double over the next five years? Will growth be driven mainly by volume, price, or new businesses?4/10

  A

  Conclusion first: it can barely double, but the quality is mixed, and it first has to climb out of negative growth. This is “moderate growth while slowing,” far from the high-speed, organic, sustainable nonlinear scaling Baillie Gifford LTGG looks for. Do the math: SDGR needs to grow from USD 255.9 million in FY2025 to more than USD 500 million in five years, requiring about +14.3% annualized growth. Its steadiest leg, software ACV, has a medium-term target of exactly +10–15%. In other words, even if software compounds steadily at the high end of the target and the lumpy drug discovery leg does not drag, a five-year doubling only “scrapes over the line,” with no margin. Worse, the starting point is not flat ground but a downhill slope: stockanalysis consensus has FY2026 revenue at USD 239 million (-6.46%), and FY2027 only back to USD 270 million (+12.77%). Baillie Gifford asks why the market has not discovered a great growth stock that can rise 5 times over the next decade. SDGR’s real script is “digest a high base with one year of negative growth, then return to a speed that just barely supports doubling.” Those are not the same order of magnitude.

  Driver breakdown: mainly “volume”; price and new businesses are not reliable engines. Broken down by the three legs:

  Growth driver	Five-year contribution assessment	Revenue quality
  Software ACV (mainly volume)	Main engine; +10–15%/year, five-year compound growth of about +60–100%	High-quality recurring revenue (subscription model, predictable, 19 of the top 20 pharma companies)
  Drug discovery (milestones / deferrals)	High elasticity but cannot be scheduled; FY26 guidance only $55–65M	Lumpy pulse, driven by accounting recognition, not suitable for linear extrapolation
  Equity portfolio (Ajax 5.8%, etc.)	Pure upside option, not included in guidance	One-off monetization, definitely not recurring revenue

  Growth in the software leg is essentially volume-driven: new customer seats, expansion within existing customers, and the agentic AI product Bunsen, planned for early access in summer 2026, aims to “expand the user base + raise platform utilization.” This is a healthy scaling logic, but the speed is moderate: the report notes ACV growth has slipped from 20%+ in earlier years to +4% in FY2025, and only “returns” to +10–15% in FY2026 guidance (Q1'26 ACV $28M, +12% YoY, full-year ACV guidance $218–228M). Price (price increases / higher average contract value) is not an explicit standalone lever. Management’s narrative is “sustainable 10–15% ACV growth,” not price increases as a separate source of growth. The move to hosted licensing has actually compressed software gross margin from 80% to 69% in Q1'26, which is “trading margin for delivery model,” not “raising revenue through price.” New businesses (pipeline milestones / equity) should not be counted in the numerator of “sustainable doubling” at all: FY2026 drug discovery guidance is only $55–65M, and the +124% surge in Q1'26 was acknowledged by the company as mainly “accelerated recognition of deferred revenue + one-time release from termination of a collaboration project.” That is an accounting pulse, not operating scale-up. Equity monetization, such as a 5.8% stake in Ajax pending Eli Lilly’s acquisition for up to $2.3 billion, is not included in guidance and is a pure option.

  After stripping out pulses, the confidence in “high-quality recurring revenue” doubling is actually weaker. If we strictly count only the recurring software leg, FY2025 software revenue was about USD 199.5 million (ACV USD 198.5 million). At +10–15% compound growth, that reaches about USD 320–400 million in five years. Software alone will not double in five years. To reach “companywide USD 256 million → USD 500 million,” SDGR must add incremental drug discovery and potential collaboration milestones, precisely the worst-quality and least schedulable revenue. Put differently, the closer SDGR gets to “doubling,” the more it depends on lumpy pulses, exactly the opposite of Baillie Gifford’s preference for organic, sustainable growth without accounting volatility. This is also why the report marks FY2026 consensus as negative growth and repeatedly emphasizes that “the financial statements are never clean.”

  Horizontal comparison 1: pure software peers are also slowing collectively, which shows this is not SDGR’s execution issue alone, but the category’s growth ceiling. Put SDGR’s software leg in a comparable pure software frame: Certara (CERT) FY2026 revenue guidance is $395–405M, with organic growth excluding divested businesses of only 0–4%; Simulations Plus (SLP) FY2026 revenue guidance is $79–82M, only flat to +4% YoY. Against that backdrop, SDGR’s software target of +10–15% actually looks “relatively fast” within the biosimulation software niche, and honesty requires giving it credit for that. But the flip side is that this is a mature niche market with structural growth only in the single digits to low teens. The “undiscovered nonlinear growth that can support 5 times over ten years” that Baillie Gifford seeks has little native soil in this category. Whether SDGR software can deliver the +10–15% target remains unproven; FY2025 actual growth was only +4%.

  Horizontal comparison 2: against LUNR, the space stock in the same framework, it becomes immediately clear what “real volume ramp” looks like. For another high-risk name rated “Watch/Hold,” Intuitive Machines (LUNR) had 2025 revenue of USD 210 million and 2026 guidance that jumps directly to $900M–$1B, multiplying its scale within one year (though largely driven by the Lanteris acquisition + NASA/defense long-contract backlog, with separate questions around quality). That is what a “bending revenue curve” looks like: moving directly from the USD 200 million range toward the USD 1 billion range. SDGR starts in the USD 200 million range, needs a full five years, and first goes through one year of negative growth, just to barely reach USD 500 million. Put the two revenue curves side by side, and through a Baillie Gifford lens it is obvious which one looks more like “a growth stock that can rise 5 times over the next decade”: SDGR’s curve is gently rising after first stepping down, while LUNR’s curve is a steep leap. This is not an endorsement of LUNR’s quality; its unit economics and acquisition-driven character are separate questions. But as a reference for ramp speed, it sharply highlights SDGR’s moderate growth.

  Honest landing point: on Question 2, SDGR cannot give the answer Baillie Gifford wants to hear. A five-year doubling is not impossible, but it requires three non-trivial conditions to hold at the same time: ① first turn FY2026’s -6.5% negative growth positive; ② keep software ACV steadily compounding at the upper end of +10–15% rather than FY2025’s actual +4%; ③ have lumpy pulses from drug discovery / equity milestones contribute incremental revenue within the window. This is a “scrape over the line, fill the gap with pulses, and start from negative growth” doubling, not a “volume and price both rising, organic acceleration, faster later” doubling. The main growth engine is software volume, healthy but moderate; price is not a standalone lever; new businesses are unschedulable options and accounting pulses. For the Baillie Gifford LTGG framework, this dimension is clearly weaker: it is a slowing company supported by a downside floor (software + net cash), not an accelerating growth engine. That is fully consistent with the report’s “Hold” rating and its note that FY2026 consensus revenue is negative growth.

  Jun 5, 2026
• QFive years from now, what will take over as the next growth engine? Does this “second curve” exist today?4/10

  A

  Conclusion first: Schrödinger’s “second curve” does exist, and it is clearly stronger than a pure option. It is built on a genuinely functioning software platform and supported by a ten-year history of repeated equity monetization. But today it is still a basket of “real but unproven options,” not a visible engine ready to take over. None of the three candidates can provide a certain baton when the first curve, software licensing, slows: Bunsen has not scaled, the owned pipeline has shrunk to two early-stage assets, and the size and timing of equity monetization cannot be scheduled. This is the growth-side reason for “Hold” rather than “Buy”: there is real asset support on the downside, but no visible growth engine that can be put on the calendar for upside.

  First define the issue: Schrödinger’s first curve is software licensing, with annual contract value (ACV) of about $198.5 million and coverage of 19 of the top 20 pharma companies. It is a real, sticky “pick-and-shovel” business. But this engine is slowing: ACV growth has slipped from 20%+ in earlier years to a medium-term target of 10–15%; software gross margin has been compressed from 80%+ to 69% by the hosted transition; FY2026 consensus revenue is even negative at -6.5%. So “whether a second curve exists today” is not a nice-to-have for Schrödinger, but the key to whether valuation can rerate. When the ballast no longer delivers high growth, the market needs to see the next source of growth. Examine the three candidates one by one:

  Candidate 1: Bunsen (agentic AI), a real option built on a real platform, but today only an “unignited engine.” Bunsen is Schrödinger’s core new narrative for 2026. StockTitan reported it as an agentic AI “co-scientist” that can autonomously execute complex molecular discovery workflows, planned for early access in summer 2026, with the company’s materials science and therapeutics teams already using it internally to improve productivity. Its growth logic is clear: combine physical ground truth with frontier AI, automate complex workflows, lower the barrier for non-expert users, thereby expanding the user base, raising platform utilization, and driving ACV. Among the three legs, it is the only baton candidate directly grafted onto the first curve (software) with a clear commercial transmission path, making it much stronger than a pure option. But honestly, as of 2026-06, it had not even formally opened early access, and there is no evidence of scale, pricing, or incremental ACV. It is a “real second-curve embryo,” but still at the “option” stage rather than the “engine” stage. Whether it can turn utilization into measurable ACV growth will only become clear over the next 12–24 months.

  Candidate 2: owned / partnered drug pipeline, downgraded from “replacement engine” to “option.” This leg has been contracting, not expanding, over the past two years. After SGR-2921 was halted in 2025-08 following two patient deaths, the owned clinical pipeline shrank to only two early-stage assets, SGR-1505 and SGR-3515, and both have shifted toward “finding partners for mid/late-stage development”. SGR-1505 recorded a 100% response rate in Waldenström macroglobulinemia, and SGR-3515 recorded a 65% disease control rate, so the data are not weak. But the narrative of self-developing a blockbuster is receding, and Schrödinger is actively handing them to partners to advance. Together with a collaboration portfolio with roughly USD 5 billion in cumulative milestone potential and 15–16 programs with future economics, this leg still has elasticity. But the path of “developing a major drug itself to take over from software” has retreated from a visible target into a low-probability, long-duration lottery ticket. It is a real option, not a visible replacement engine.

  Candidate 3: equity portfolio monetization, real cash flow repeatedly proven, but impossible to schedule. This is Schrödinger’s most distinctive component and the one SOTP bulls value most: Nimbus contributed about $147.3M and Morphic about $47.6M; since 2016, cumulative cash realization has been about USD 700 million. The “co-create, get acquired, receive proceeds” model has been repeatedly proven over ten years. The current live catalyst is Eli Lilly’s April 2026 announcement to acquire Ajax for up to $2.3 billion (milestone-based consideration), with Schrödinger owning about a 5.8% stake. But the monetization amount has not been disclosed and is not included in 2026 financial guidance. The character of this leg is clear: it can repeatedly deliver, which is hard evidence that makes it far stronger than a pure option, but the size and timing of each realization cannot be predicted or put on a calendar. It can extend the cash runway and deliver surprises, but it cannot be modeled as a smooth, predictable growth curve. It is an option that “can win repeatedly but with random draw dates,” not a baton engine.

  Horizontal comparison clarifies the quality of the “second curve.” Compared with SIDU (Sidus Space), where the first curve itself is shrinking and breaking and the supposed second curve is a pure lottery ticket, Schrödinger is fundamentally different. Its first curve, software, is slowing but not broken; it remains a real business with about USD 200 million in revenue, 70% gross margin, and 19/20 top pharma customers. All three candidate curves have a real platform or historical proof behind them; they are not stories conjured from nothing. But compared with LUNR (Intuitive Machines), Schrödinger’s weakness is visible: LUNR’s second curve is built on a first curve of about USD 200 million that is still growing healthily, and that first curve itself is funding and paving the way for the second curve. Schrödinger’s first curve is slowing, and none of the three second-curve candidates has already scaled. So it sits between SIDU and LUNR: far better than “betting a pure option on top of a shrinking first curve,” but short of “a second curve emerging from a healthy-growing first curve.”

  Honest landing point: this is a basket of “real but unproven options,” not a visible baton engine. Taken together, Bunsen is built on a real software platform with a clear transmission path but has not yet scaled; equity monetization has a hard ten-year history but cannot be scheduled; the owned pipeline is shrinking and has been downgraded to a low-probability lottery ticket. Together they form an option basket that is far more solid than pure options, yet still unproven: the platform is real, the history is real, and the options are real, but no line item can yet take over from slowing software with certainty or be inserted into the next few quarters’ results. That is the answer to Question 3 for Schrödinger: the second curve exists and is not empty, but what it delivers today is “downside support + upside options,” not “visible replacement growth.” To upgrade this basket of options into a true baton engine, Bunsen must produce measurable ACV after early access, or a major pipeline partnership must close, or Ajax-style monetization must continue to happen. Until then, the growth side cannot provide the visible handoff required for “Buy”; it remains “Hold, wait for ignition.”

  Jun 5, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?5/10

  A

  Conclusion first: Schrödinger’s core competitive advantage is real and verifiable: a physics-first foundation (FEP+ as the free-energy perturbation gold standard), layered with workflow stickiness across 19/20 large pharma companies, 30+ years of algorithmic accumulation, NVIDIA compute collaboration, and the scarce positioning of a “physical truth anchor” in the AI era. It is clearly different from AI drug discovery peers that are mostly narrative. But the honest judgment on whether the moat will widen or narrow over the next three to five years is this: the technical moat may widen slightly as AI increasingly needs physical calibration, while at the same time it is being eroded by open-source AI (Boltz-2) on speed/cost, by large pharma building internal compute, and by the fact that the moat still protects “technical leadership,” not “earning power.” The net result is closer to “width roughly stable, profit width unproven.” This is SDGR’s strongest dimension, but far from “unassailable,” and consistent with a “Hold” rating.

  Core competitive advantage: four pillars, with varying substance

  The report defines three pillars of the moat. After verification, I refine them into four and mark the “hardness” of each:

  Pillar	Content	Hardness assessment
  ① Physical interpretability	FEP+ calculates binding free energy based on quantum mechanics / molecular dynamics, is treated by the industry as a “gold standard” for affinity prediction, and can extrapolate into new chemical space	Hardest, a real technical barrier, supported by objective benchmarks below
  ② Customer stickiness + compute	Used by 19 of the top 20 pharma companies, embedded in R&D workflows with high switching costs, 30+ years of accumulation since 1990, NVIDIA collaboration	Hard, but “used” does not equal “irreplaceable”; see erosion forces
  ③ AI truth anchor	Physical ground truth + frontier AI integration; AlphaFold3 is strong in static structures, while conformations / affinity ranking require physics to fill gaps	The trend helps, but this is “narrative + trend” and needs execution
  ④ Commercial monetization moat	(bull argument in the report) software ballast + 30+ year brand	Softest; this is precisely the gap, not an advantage, because software still does not make the company independently profitable

  The key honest point: ① and ② are real barriers; ④ is a weakness packaged as a strength. The moat protects “technical position,” and the report has already punctured the issue that “technical leadership has not yet proven it can reliably convert into commercial profit; software is not independently profitable, and the owned pipeline has not produced a major drug.” The widening/narrowing judgment below must separate “technical width” from “profit width.”

  Forces widening the moat: the AI era’s “need for physical calibration” is a real tailwind

  The strongest objective evidence comes from competitors’ behavior, not Schrödinger’s own messaging. Recursion (RXRX), used in the report as the “pure AI black box” comparison, and MIT open-sourced Boltz-2, claiming affinity prediction “near the accuracy of physics-based FEP but 1000 times faster”. Yet in October 2025, Recursion itself released a pipeline combining Boltz with “absolute binding free energies,” i.e. a physical method. Even the most aggressive AI challenger is moving back to “add physics.” Academia confirms the same pattern: AlphaFold3 is good at static structures but usually provides only a single conformation and lacks binding-state flexibility, requiring further refinement; mainstream practice is moving toward a hybrid pipeline of “AI fast generation → physics-based refinement and ranking → experimental validation.” This means the more broadly AI is adopted, the more demand rises for calibration by a “physical truth anchor.” Schrödinger’s pillar ① benefits from this trend. Add Bunsen, the new agentic AI co-scientist with summer 2026 early access, intended to automate complex discovery workflows, bring tools to non-expert users, and raise platform utilization, plus switching costs that accumulate with usage time, and these are genuine widening forces.

  Forces narrowing the moat: open-source AI speed/cost pressure + large pharma internal builds + profit width still unproven

  But the countervailing erosion is also hard and more urgent:

  First, open-source AI crushes FEP+ on speed/cost in “good enough” use cases. Boltz-2 is fully open-sourced under the MIT license (model, weights, and training pipeline all commercially usable), and can complete one affinity prediction on a single consumer GPU in about 18 seconds, while the same task using physical methods takes hours or even days, with per-molecule cost often above USD 100. In many early-stage “rough screening, no need for 0.x kcal/mol precision” scenarios, a free and 1000-times-faster open-source tool is enough to eat into the paid rationale for FEP+. This directly weaponizes the weakness the report identifies in FEP: compute-heavy and slow.

  Second, large pharma building internally reduces external procurement. Industry data show Pfizer, Roche, AstraZeneca, and others each investing more than USD 500 million in internal AI capabilities. Schrödinger’s “19/20 large pharma users” stickiness will partially be offset as customers’ internal platforms replace part of external purchasing budgets. The report’s Pre-mortem “scenario 2” already lists “large pharma building internal computational capabilities and reducing external procurement” as a core path by which the software ballast could loosen.

  Third, and most importantly, the gap between “technical width” and “profit width” has not closed. Put Schrödinger into a horizontal matrix, and the gap is obvious:

  Company	Model	Profitability status	Moat implication
  Schrödinger SDGR	Physics + pipeline + equity hybrid	No (targeting adjusted EBITDA breakeven in 2028)	Hardest technology, unproven profit
  Certara CERT	Biosimulation software + services	Adjusted profitable, software revenue +18%, adjusted EBITDA USD 34.80 million +20%	Less “sexy” technology than SDGR, but moat has converted into profit
  Simulations Plus SLP	Pure PBPK modeling software	Yes	Same, a narrow but stable profit moat
  Recursion RXRX	AI black box → heavy pipeline	No	Speed/scale narrative, but weak unit economics and interpretability
  LUNR (same-framework comparison)	NSN lunar relay sole-source	No	Local real exclusivity (contract moat), but narrow scope and dependent on external policy

  Certara/SLP prove that the “computational drug discovery software” business can be profitable. Their moats are less technically elevated than Schrödinger’s, but they have already converted moat into earnings. Schrödinger has the strongest technology, yet is dragged down by owned-pipeline spending, and software still cannot make the company independently profitable. Put differently, Schrödinger leads peers on “technical width,” but on the “profit width” that actually determines shareholder returns, it trails Certara, which should have been an easier rival to beat. Compared with LUNR in the same framework, LUNR is a “narrow but real” contractual monopoly (NSN relay local sole-source), while Schrödinger is “wide but soft” technical leadership. Neither has yet converted the moat into stable profit.

  Objective benchmark for the moat: FEP+’s technical lead still holds today, but the advantage is narrowing

  To avoid distortion on either side, use the hardest objective anchor: on the recognized 4-target FEP+ benchmark (CDK2/TYK2/JNK1/p38), open-source Boltz-2 has Pearson correlation of about 0.66, tied with open-source OpenFE, and still below commercial FEP+ at 0.78; moreover, co-folding methods like Boltz-2 cannot adjust the binding pocket to tolerate initial conformation errors the way FEP can, limiting quantitative ranking accuracy. In other words, as of mid-2026, FEP+ has not been replaced in the core function of “quantitative accuracy,” and the technical moat still stands. But the 0.78 versus 0.66 gap is a trade-off between “more accurate” and “free and 1000 times faster.” As open-source models iterate, the “pricing-power premium” of this moat will keep thinning. Technical leadership remains, but both the margin of leadership and the chargeable thickness are narrowing.

  Final judgment

  Over the next three to five years, Schrödinger’s moat will probably be stable or even slightly wider in “technical width” because of AI-era demand for a truth anchor (hybrid pipelines become standard, Bunsen expands usage, switching costs accumulate, and even competitors add physics). But it has still not proven widening in “profit width”, while simultaneously facing open-source AI’s speed/cost pressure and large pharma internal builds. This is SDGR’s real and relatively strongest dimension. It is not an empty shell like BenevolentAI/Exscientia where narrative exceeds delivery; the physical gold standard, top-tier customers, and repeatable equity monetization (about USD 700 million cumulatively from Nimbus/Morphic/Ajax) are all tangible moat evidence. But “the moat is real” and “the moat makes money” are two different things: Schrödinger has proven the former, not yet the latter. Baillie Gifford is looking for a growth stock that can rise 5 times over the next decade and whose greatness the market has not recognized. Schrödinger’s moat is real enough, but it remains stuck in a state of “technical width, narrow profit width, no catalyst.” That warrants “Hold,” not “Buy,” and falls short of “unassailable.”

  Jun 5, 2026
• QIf the core business is disrupted, does it have the DNA to reinvent itself? How does it handle mistakes and bad news?5/10

  A

  Conclusion first: SDGR has a real posture of reinvention, and its handling of bad news is in the honest tier among biotech peers. This is the substantive difference between it and governance-red-flag names. But stay clear-eyed: this reinvention is a defensive transition after pipeline setbacks, shrinking back toward high-gross-margin software, supported by a thick USD 300 million net-cash cushion rather than self-funded cash generation, and it has never been tested by a “near-death and rebirth” cycle. So on the dimension of “reinvention DNA,” it deserves medium-high, not top-tier.

  First make the premise explicit: Baillie Gifford is not merely asking “can it choose another path if a pipeline fails,” but the sharper question: if SDGR’s foundational methodology, physics-based computation (FEP+), is itself disrupted by generative AI or faster data-driven methods, does it have the DNA to reinvent itself? That is the real core-business disruption scenario. On this question, SDGR’s answer is relatively positive. It has not clung dogmatically to physics; it has actively repositioned physics as the “truth anchor” in the AI era, using physical simulation to generate training data for AI models, as autonomous driving and weather forecasting use simulation data to train models, rewriting “AI will disrupt me” into “AI needs my ground truth.” Together with Bunsen, the agentic AI product planned for summer 2026 early access, a “co-scientist” that autonomously executes complex molecular discovery workflows and is meant to bring in both expert and non-expert users while lifting platform utilization, this is a posture of embracing rather than resisting the AI wave. The report’s “truth anchor” framing is consistent with company messaging, not an overstatement.

  The more decisive reinvention is strategic contraction at the business-model level. In May 2025, the company decided to stop independently developing SGR-1505 and SGR-3515 after Phase 1 and cut headcount by 7%, pulling resources back from “capital-heavy self-developed blockbuster drugs” toward “software licensing + partnered / preclinical discovery,” the two legs it believes have the best synergy and economics. For the remaining two assets, it shifted toward “finding mid/late-stage partners” (management explicitly said it was shopping the assets, seeking USD 50 million–100 million upfront plus milestones). Add the software hosted/cloud transition, which temporarily compresses gross margin from 80% to 69% with a target of returning to high-70s in 2028, and this is a clear self-negating redirection: it admits that “making major drugs by itself” is too expensive and biology-constrained, and retreats to the higher-gross-margin model of pick-and-shovel software plus equity economics. The report’s view that strategy is shifting from “heavy self-development” toward “high-gross-margin software + partnered economics + AI truth anchor” matches the outside facts.

  What really earns it credit is its attitude toward bad news, the strongest part of SDGR’s corporate constitution. The handling of SGR-2921 (CDC7 inhibitor) is a textbook case. On August 14, 2025, after two AML patient deaths occurred during Phase 1 dose escalation and the company acknowledged that SGR-2921 was “considered to have contributed” to those deaths, it did not polish the story, did not delay, and announced the termination of the entire program the same day. The chief medical officer said, “patient safety is our highest priority… although early clinical activity had been observed, we believe this is the right decision for patients.” Faced with a situation of very high sunk cost, early efficacy signal, and the temptation to keep looking for combination approaches, the company chose to stop decisively and disclose transparently. The stock’s -16% move that day was the market’s immediate price for that honesty. This is real evidence of Baillie Gifford’s idea that reinvention requires candid handling of bad news.

  More broadly, the report’s own framing also supports the honesty of SDGR’s information environment: FY2026 revenue consensus is negative (-6.5%); software ACV growth has slowed from 20%+ to 10–15%; gross margin is compressed to 69% by the hosted transition; Gates Foundation contribution revenue has ended; the owned pipeline has shrunk to only 2 assets. These bad facts are all disclosed directly by the company and the report, not buried in footnotes. This contrasts sharply with governance-red-flag names, such as microcaps with ineffective disclosure controls, auditor issues, or related-party transactions: SDGR is a company that tells you bad news itself, not one where investors have to dig hidden mines out of the notes. This transparency is the deepest and hardest-to-fake precondition for “reinvention DNA” in the Baillie Gifford framework.

  But an honest score must be capped by three points: First, this reinvention is “passive contraction,” a defensive retreat after pipeline setbacks, not an offensive rebirth. The company is returning to the software comfort zone after the self-developed blockbuster story cracked; the direction is right, but it is more “conceding” than “breaking through.” Second, it has no record of “surviving a near-death crisis and being reborn”. SDGR has never truly approached existential danger and climbed out through self-generated cash flow. Its composure comes from a thick USD 300 million net-cash cushion (about USD 600 million of cash realized over the past five years through upfront payments / company sales / milestones, USD 700 million under the report’s framing), and survival has depended on equity monetizations such as Nimbus/Morphic/Ajax, not on the software business already being profitable by itself. Third, on the ultimate question of “methodology disruption,” the truth-anchor story remains more narrative than proven: Bunsen has not scaled, and the physics approach still has the weaknesses of being compute-heavy and slow relative to AI screening. Whether “AI needs me” can become ACV rather than slides remains to be proven.

  In sum, SDGR’s profile on “reinvention + handling bad news” is clear: it has a real reinvention posture, repositioning physics as an AI truth anchor, launching Bunsen to embrace AI, and decisively contracting capital-heavy self-development back toward high-gross-margin software; it also handles bad news honestly and transparently, stopping SGR-2921 decisively and explicitly disclosing negative growth / slowdown. This makes it materially stronger than governance-red-flag names. But the reinvention is passive contraction, untested by a near-death self-funded recovery, and supported by a thick cash cushion rather than self-generated cash flow. This is a medium-high answer: it has the scarce prerequisite Baillie Gifford values, candid handling of bad news, but lacks the hard evidence of reinventing at the edge of a cliff and emerging through its own cash generation. So it is not top-tier.

  Jun 5, 2026
• QDoes management, especially the founder, have a long-term view and deep alignment with the company? Is it willing to sacrifice current profit for five to ten years from now?4/10

  A

  Conclusion first: medium. The governance structure is far cleaner than SIDU-type names, with no super-voting control, no related-party transactions, and no destructive dilution. But the “deep owner-operator alignment” that Baillie Gifford values most is clearly weaker than at LUNR: founder Friesner has stepped back to a scientific-advisor role and no longer runs the company, the current CEO is a professional-manager scientist with direct ownership of only about 0.5% and planned selling, while the Gates Foundation and David Shaw, the two “godfather-level patient capital” holders, have either previously cut holdings or ended funding. The company has indeed sacrificed current profit for 5–10 years out, but we must distinguish active long-termism from being unprofitable by necessity.

  First: are the founder and management still heavily exposed to downside? Alignment is structurally weakening. The core of the Baillie Gifford template is “founder owns heavily for the long term and is tightly aligned with the company.” SDGR is moving in the opposite direction, toward “de-founderization.” As the report states, founder Richard Friesner has become co-founder / chair of the scientific advisory board / director and no longer handles day-to-day operations; day-to-day leadership has been with Dr. Ramy Farid, CEO since 2017. Friesner still owns about 1.23 million shares as a director (about USD 19 million at $15.85), so real money remains at stake, but he is no longer the operator “sacrificing everything for the company.” More notable is CEO Farid’s own alignment: after a series of sales, he currently directly owns about 330,000 shares (about 130,000 of which are unvested RSUs), only about 0.5% of total shares. In April 2026, he again exercised and sold 86,000 shares under a 10b5-1 plan. This is the profile of a scientist professional manager, not the owner-operator Baillie Gifford prefers whose personal net worth is pressed into the company. Horizontally, LUNR’s founder truly owns about 33.5%; SDGR’s core management economic alignment is not in the same league.

  Second: major shareholder structure. The aura of “patient capital” has visibly faded and creates long-term supply pressure. One of SDGR’s former selling points was its shareholder register: the Gates Foundation Trust + David Shaw-related entities. That aura is fading, and the framing must be precise. The Gates Foundation Trust sold 2 million shares at a high level around $63.50 in 2020, realizing about USD 127 million, reducing its stake from about 10 million shares to about 7 million. Since then, it has been broadly stable at 6.98 million shares (about 9.3–9.5%), still around that level as of Q1 2026. So the more precise statement is: Gates’ “large reduction” mainly happened years ago at high prices; today it is a large overhanging stock position that has been broadly unchanged over the past year. It has neither continued to slash nor signaled renewed accumulation. Add the report’s fact that Gates-funded contribution revenue has ended (this revenue fell from USD 4.3 million to USD 0.1 million in Q1 2026), and the signaling value of “godfather endorsement” has weakened. On David Shaw, a common misunderstanding needs clarification: D.E. Shaw hedge fund 13F filings show SDGR holdings at 0, but that is the fund-entity view. David E. Shaw personally / related entities still hold about 6.2 million shares, about 8.3%, still one of the top three shareholders. In other words, the two “godfathers” have not exited, but both have moved from “believers adding capital” to “large existing holders not adding.” That is a negative, not a positive, for Baillie Gifford’s emphasis on long-term capital continuing to stand behind the company.

  Third: willingness to sacrifice current profit for 5–10 years out. The posture is real, but distinguish “active sacrifice” from “forced losses.” This is where SDGR earns credit. The company continues to post large losses (net loss 2024 -187.1 million → 2025 -103.3 million → 2026 Q1 -60 million), and a substantial portion comes from internal R&D spending on owned pipeline assets. Despite knowing this burns cash and displeases the market in the short term, it still invests cash into drug pipelines that may only pay off 5–10 years later, and has put adjusted EBITDA breakeven only in 2028 into guidance. This is genuinely a long-termist posture of sacrificing current profit for the future, and it fits the Baillie Gifford direction. But it must be discounted honestly: first, the high-gross-margin software business could likely be profitable on its own, and it is the self-developed pipeline that drags the company into sustained losses, so this is an active capital allocation choice (positive). Second, after SGR-2921 was stopped because of two patient deaths, the owned pipeline has shrunk to 2 assets, and both have shifted toward “finding partners for later-stage development.” That means the long-term sacrifice story is being invested in while also being trimmed by reality; it is not a full-bore bet driven by a powerful founder’s will.

  Fourth: governance red-flag check. Clean, and that is a real plus. In sharp contrast with SIDU, the negative example in this framework, SDGR has no dual-class super-voting structure allowing a founder to control the company with tiny economics, no related-party transactions equal to nearly half of revenue, and no hundredfold destructive dilution over two years. The board chair is outside director Michael Lynton, CEO and chair are separated, and institutional ownership is normal, with BlackRock around 13%, Vanguard around 5%, and other index / active funds. The only forward-looking risk to watch is potential dilution. The company targets profitability only in 2028, and if pipeline progress runs faster than the cash runway, it could still use ATM issuance like peers RLAY and ABCL. But that is a “risk not yet happened,” not an already realized governance blemish.

  Combined landing point (Baillie Gifford view): medium. Put the pieces together: it has sacrificed profit for the long term (+), and governance is clean with no red flags (+), but the key Baillie Gifford yardstick of “founder / management heavily owns and shares downside” is clearly weaker. The founder has stepped back to advisor, CEO economic alignment is only about 0.5% and he is selling, the two godfather-level capital holders have shifted from “believers adding” to “large holders not adding,” and Gates funding has ended (−). Its governance quality is far better than SIDU (economic interest <1% with super-voting control + 47% related-party transactions + hundredfold dilution), but owner-operator alignment is far weaker than LUNR (founder truly owns about 33.5%). In one sentence: this is a company with respectable governance and a real long-term investment posture, but it has lost the soul of a forceful founder-operator, and top-tier long-term capital is no longer adding. In Baillie Gifford’s paradigm of “walking alongside exceptional management for the long term,” it is “acceptable,” not “exciting.”

  Jun 5, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable and not dependent on harming society or exploiting regulation?5/10

  A

  Conclusion first: customers would genuinely miss it, but what they would miss is the capability represented by “FEP+ as the gold-standard shovel,” not necessarily “Schrödinger and only Schrödinger.” Its growth model is half healthy and half life support: subscription software is a sustainable good business; the self-developed pipeline depends on cash burn plus financing; equity monetization cannot be scheduled; overall, the company is supported by USD 300 million of net cash rather than fully self-funded cash generation. But one thing can be stated firmly: its field, drug discovery, is inherently legitimate, with no negative social externality, no regulatory arbitrage, and no dependence on harming anyone for growth. This is where SDGR is significantly stronger than SIDU (demand concentrated in related parties, completely substitutable supply, no sole-source lock-in) and is the most solid relative strength of this Hold-rated company.

  Start with indispensability: it is real, and fairly high. The report’s fact chain is hard: FEP+ (free-energy perturbation) is treated by the industry as the gold standard for predicting how tightly small molecules bind to targets; 19 of the world’s top 20 pharma companies use it; once embedded in pharma R&D workflows, it becomes a mission-critical tool with very high switching costs. This is not marketing copy. Academia and vendors describe alchemical free-energy calculation as a “gold standard” for enhancing experimentally driven drug discovery, and Schrödinger’s own FEP+ product line centers on computational accuracy matching experimental methods across broad chemical space. So if it disappeared tomorrow, 19 large pharma companies would not shrug. They would either lose a proven prediction engine already trained on their project data, or be forced to rebuild an equally trusted and interpretable physical computation capability. Under the report’s framing, the latter is built from “more than thirty years of algorithmic accumulation + NVIDIA compute collaboration,” and cannot be copied quickly. They would genuinely miss it. That point holds.

  But the honest assessment is that “missing this capability” does not fully equal “only it can provide it”; substitutes are increasing. The soft point must be made explicit: ① large pharma is building internally. Eli Lilly and NVIDIA are co-building a USD 1 billion co-innovation lab, using NVIDIA BioNeMo and Clara foundation models to build proprietary computational pipelines. Top customers are both SDGR’s funders and developing their own teeth. ② Generative AI is approaching FEP+ accuracy. A 2025 study fine-tuned AlphaFold3-like models (Protenix) on bioactivity data to create AlphaRank, claiming “FEP+-comparable accuracy” on standard benchmarks with far lower compute; MIT/Recursion released Boltz-2 in June, jointly predicting structure and affinity and claiming to be “1000 times” faster than physics-based FEP. This maps directly to the report’s weakness for FEP, heavy compute and slow speed (about 10 hours for relative binding free energy calculation of a single ligand, and $10+ in cloud compute per affinity value). Conclusion: what customers cannot do without is the capability of accurately predicting binding affinity. SDGR is currently the best and most trusted provider, but not the only one. It has near-gold-standard leadership, but not LUNR-style sole-source exclusivity.

  Compared with LUNR in the same framework, this is where the gap lies. Intuitive Machines’ deep-space network (NSN) is structurally sole-source within the NASA system; no second provider can offer equivalent deep-space tracking and communication. FEP+ is a de facto-standard form of indispensability: technical leadership + workflow embedding + switching costs build a high wall, but outside the wall AlphaFold3-family tools, Boltz-2, open-source tools, and customer internal builds are all approaching. So SDGR’s indispensability is stronger than SIDU’s, which has no lock-in at all, but weaker than LUNR’s institutional exclusivity. That is the honest middle position. It is also worth noting that there is currently no evidence of customer loss due to AI substitution: the company’s 2025 software revenue still grew 11%, with “high customer retention and strong customer engagement”, ACV is moving from USD 198 million toward USD 218–228 million, and the company targets “sustainable 10–15% ACV growth.” The real pressure is growth slowing from 20%+ to 10–15% and gross margin falling from 80%+ to 69%, a “slowdown,” not “customer churn.” That actually supports the idea that customers do miss it: they have not left, but marginal expansion is slowing.

  Now sustainability of growth: the three legs have very different health. This is the second layer of the question and must be separated honestly. ① Subscription software (healthy and sustainable): about USD 200 million a year, subscription-based, predictable, and sticky. This is textbook sustainable good-business cash flow and the ballast of SDGR’s value. ② Owned pipeline (kept alive by cash burn + financing): the report is clear that the company has continued large losses (FY25 net loss USD 103 million), the owned pipeline has shrunk to 2 assets and both are seeking partners, and cash is being burned on R&D with a target of adjusted EBITDA breakeven only in 2028. This leg is not self-funded; it is using cash cushion to buy time. ③ Equity monetization (one-off, unschedulable): Nimbus/Morphic/Ajax have cumulatively realized about USD 700 million, which has indeed been repeatedly proven, but the report itself emphasizes that “amounts and timing cannot be predicted” and are “not included in guidance.” This is upside option value, not reliable operating cash flow. Together, growth is sustained by “healthy software subscriptions + a thick cash cushion (USD 300 million net cash supporting 2–3 years),” not by fully self-generated cash flow. The growth model is half sustainable and half balance-sheet-supported until the profitability inflection.

  The most important positive: the field is legitimate, with zero social negative externality and zero regulatory arbitrage. This is one of the cleanest aspects of SDGR versus many growth stocks and deserves explicit credit: it works on drug discovery, helping pharma companies find disease-treating molecules faster and cheaper. That is socially positive by nature. There is no element of growing by misleading users, polluting the environment, exploiting regulatory loopholes, or harming third parties. Its revenue comes from pharma paying for R&D efficiency (subscriptions), from drugs it discovers actually working (pipeline / royalties), and from co-created companies being acquired (equity). Every leg is built on the legitimate act of “getting drugs made.” Compared with the kinds of growth this framework should be wary of, such as regulatory arbitrage or externalizing social costs, SDGR is almost beyond criticism on this dimension. This is the fundamental split from SIDU: SIDU’s demand is related-party concentrated, its supply fully substitutable, and it has no sole-source status; SDGR has real demand (pharma need) + real barriers (gold standard + switching costs) + a legitimate field.

  Honest landing point: customers would genuinely miss it, and the field is legitimate and sustainable. This is SDGR’s real relative strength. But “indispensable” means “hard-to-replace leader,” not “irreplaceable monopoly,” and overall growth still depends on a cash cushion rather than complete self-funding. In one sentence: a strength, but not top-tier exclusivity. On “how much customers would miss it tomorrow + sustainability,” its score should be clearly above SIDU, whose supply is substitutable and demand related-party-driven, but a notch below LUNR’s sole-source exclusivity. FEP+’s moat is a de facto-standard high wall, not an institutional sole source. On growth sustainability, “healthy subscription software” is valuable, but “using a thick cash cushion to survive until a 2028 profitability target” reduces the quality of full self-funding. This supports the “solid downside floor” in a Hold rating, but not top-tier “irreplaceable without thinking.”

  Jun 5, 2026
• QWhat are the unit economics of this business (gross margin, incremental returns)? Do they improve or worsen with scale? Where does the money it earns go?5/10

  A

  Conclusion first: unit economics are “strong in software, unproven at the company level,” medium-high rather than top-tier. Broken apart, SDGR’s software unit economics are textbook high-quality SaaS: historical gross margin above 80%, subscription recurring revenue, very low marginal cost, and better unit economics as scale grows. This is far stronger than pure cash-burning SIDU (gross margin -168%→-292%, deeply negative, negative scale effects, reliant on financing) and is a real bright spot. But companywide unit economics remain negative because owned-pipeline R&D spending consumes them: the company continues to post large losses (net loss 2024 -187.1 million → 2025 -103.3 million → Q1'26 -60 million), software gross margin is temporarily moving in the wrong direction (down to 69% in Q1'26), and the profitability target is not until 2028. The compounding machine Baillie Gifford prefers, where scale makes FCF snowball, has only been proven in “the software half” of SDGR; the overall compounding machine has not formed.

  ① Software unit economics: high-quality SaaS, but moving downward in the short term. The software leg has historical gross margin above 80%, ACV of about USD 198.5 million, coverage of 19 of the top 20 pharma companies, and predictable subscriptions. Marginal customers cost very little, a typical positive operating-leverage model where “unit economics improve with scale.” But it is currently in transition pain: because of the shift to hosted/cloud licensing, Q1'26 software gross margin fell from about 80% a year earlier to 69%, and hosted revenue’s share of software revenue rose from 24% to 34%. This is transitional rather than structural: hosted revenue is recognized ratably over time rather than upfront, which naturally depresses reported margin during the transition, in exchange for a more predictable recurring-revenue base. Management targets about 75% hosted mix by 2028 and gross margin back to high-70s. Judgment: the transition logic is credible, as subscription transitions usually are, but the current fact must be acknowledged honestly: short-term gross margin is still moving down and the recovery is not expected until 2028.

  ② Companywide unit economics: dragged negative by internal R&D, with incremental returns kept alive by cash cushion. Combining high-gross-margin software with pipeline cash burn, the company as a whole is still deeply loss-making. Incremental revenue is not producing positive incremental profit; it is being consumed by owned-pipeline R&D. The point is that GAAP net loss includes many non-cash items, such as stock-based compensation and fair-value changes in securities, so real cash burn is lower than reported loss. Normalized operating cash consumption is about USD 100–150 million annualized, and cash + securities were about USD 402 million at year-end 2025, implying about 2.5 years of runway, with equity monetization such as Ajax potentially extending it. In other words, at the company level it is currently “negative incremental return + cash-cushion survival,” not self-funded compounding.

  Where the money goes: owned pipeline + hosted software transition + R&D. Half is high-return reinvestment, and half remains “cash burn awaiting validation.” FY2025 revenue of USD 255.9 million = software about USD 199.5 million + drug discovery USD 56.4 million. The money mainly goes to three places: (a) the software hosted transition and platform R&D, including Bunsen and other agentic AI work, intended to strengthen the high-gross-margin SaaS moat and recurring revenue, a high-return reinvestment; (b) owned-pipeline R&D, the major cash-burn item, whose return is not yet proven (SGR-2921 was terminated after two patient deaths, and the owned pipeline has shrunk to 2 assets seeking partners). So whether the spending is “worth it” is internally split: reinvestment on the software side has a clear unit-economic return, while the pipeline side remains a high-risk, lumpy, unproven option.

  Net scale effect: it depends on whether software can absorb internal R&D. Software has positive leverage (SaaS scaling), while the pipeline has negative leverage (lumpy spending, revenue driven by milestones / deferred recognition, highly uneven quarter to quarter). The net effect does not automatically improve. Companywide unit economics only turn positive if software ACV grows sustainably at 10–15%, gross margin returns to high-70s in 2028, and internal R&D burn declines as trials wind down. That is the core bet behind the 2028 adjusted EBITDA breakeven path: software compounding absorbing pipeline burn. The path is medium-credible: directionally reasonable and the runway is sufficient, but it depends on software reaccelerating and the transition finishing on schedule, both of which are still “guidance” rather than “delivered.”

  Horizontal comparison (unit economics positioning):

  Company	Model	Gross margin / unit-economics quality	Companywide profitability
  SDGR	Software + pipeline + equity	High software gross margin (80%+, temporarily compressed to 69%), overall dragged by internal R&D	No (target 2028 adjusted EBITDA breakeven)
  Certara (CERT)	Software + services	Healthy and already profitable	Yes (GAAP net income positive, adjusted EBITDA margin about 32%)
  Simulations Plus (SLP)	Pure software	Healthy gross margin (about 58–64%), but FY2025 turned to net loss because of impairment	No (FY2025 net loss, including goodwill impairment), so the “already profitable software peer” label needs discounting here
  Intuitive Machines (LUNR)	Space services	Gross margin just turned positive consecutively (Q1'25 about 11%), direction right but absolute level low	No (guides to 2026 adjusted EBITDA breakeven)

  Comparison conclusion: SDGR software’s “high-gross-margin SaaS” unit economics are stronger than LUNR’s, where gross margin has only just reached low positive levels, and stronger than SLP at the moment after impairment-driven net loss; but weaker than Certara, where software + services have already produced real companywide profit and adjusted EBITDA margin of about 32%. So the right label is “excellent software unit economics but unproven companywide profitability”: a notch above pure cash-burning, negative-scale-effect companies, but not at the level of peers that have already used SaaS leverage to deliver profits.

  Honest landing point: SDGR’s software unit economics are a genuine bright spot (high gross margin, predictable subscription revenue, SaaS positive leverage, far better than SIDU’s deeply negative gross margin, negative scale effects, and pure financing dependence), which earns it medium-high. But do not elevate it too far: the company is still persistently loss-making, software gross margin is temporarily moving down from 80% to 69%, EBITDA breakeven is only targeted for 2028, and current incremental returns are dragged negative by internal R&D. Unit economics are “software strong, companywide unproven”: high-quality SaaS under pressure + unvalidated overall cash generation, medium-high but not top-tier.

  Jun 5, 2026
• QWhat conditions must all hold for it to rise 5 times over ten years? Are those conditions realistic? What expectations are embedded in today’s share price?3/10

  A

  Conclusion first: a 5 times return over ten years (market value from about USD 1.18 billion to about USD 6.0 billion) requires a long chain of low-probability events to happen together, so the compounded probability is low, especially from the starting point of “software slowdown + FY2026 revenue decline,” which is the hardest headwind. But unlike SIDU, where the price is already decisively overdrawn, SDGR at $15.85 is only at the lower end of the report’s base range ($15–22). The market is pricing it conservatively on “software + net cash” SOTP, giving almost no credit to pipeline and equity options. So the reason “5 times” fails would be “growth engine slowing and no catalyst,” not “the price has already overpaid for the future.” There are real assets supporting downside; upside needs ignition. Honest landing point: conditions are unrealistic, but the price is not a trap; medium-weak.

  Lay out the multiplicative condition chain first. Every link must hold. To move from about USD 1.18 billion to about USD 6.0 billion, at least six things must happen together: ① software ACV must reaccelerate from the current 10–15% and sustain it long term. But 10–15% alone only produces about 2.6–4 times software revenue over ten years, not enough for 5 times market value, so it must accelerate again to 15%+; ② gross margin must recover from Q1’s 69% to high-70s, adjusted EBITDA must turn positive in 2028, and the company must move toward real GAAP profitability; ③ the pipeline, now reduced to 2 early-stage assets (SGR-1505, SGR-3515, both seeking partners), must produce at least one major drug or a large late-stage partnership; ④ the equity portfolio must continue to monetize (Ajax 5.8% realization + a new wave of co-created companies acquired); ⑤ dilution must stay low over the decade, even though the company remains loss-making and peers RLAY/ABCL have used ATM issuance; ⑥ the valuation yardstick must shift, with the market moving from today’s “value trap” lens to a “profitable growth software” lens. Treat these six as independent probabilities and multiply them, and you get “multiple medium-to-low-probability events must all happen.” That is the mathematical difficulty of a 5 times return over ten years.

  Looking at realism link by link, the hardest headwind is the first link. Sell-side consensus puts SDGR’s revenue growth over the next three years at about 13–17% CAGR, but that starts from FY2026 consensus revenue of -6.5%. An engine that is slowing and contracting in the current year has to reaccelerate over a ten-year horizon to a level that can support 5 times. That is contrary to the recent trend. Profitability is even more important: the same sell-side models still forecast SDGR loss-making over the next three years (2027 EPS about -$1.45, 2028 about -$0.64). So even “positive in 2028” remains a company target, not a market-modeled fact; and even if achieved, it is only adjusted EBITDA breakeven, still far from the real profit needed to support a USD 6.0 billion market value. Links three and four, pipeline success and equity monetization, are extensions of a “sell shovels + prospect for gold” model that has cumulatively realized about USD 700 million over the past decade, so there is evidence but no schedule. Link six, rerating, simply waits for the market to agree. None of these three links sits inside near-term company-controlled certainty.

  Ajax, the hottest catalyst, is a real option but still unrealized and conditional. Lilly’s acquisition of Ajax was announced on April 27, 2026 (not yet closed), and the consideration is structured as “up to USD 2.3 billion” in milestones. First proof-of-concept clinical data for the core asset AJ1-11095 is still expected later in 2026. SDGR’s 5.8% realization amount is undisclosed and not included in guidance. It is a clean upside option, but the cash from “up to USD 2.3 billion × milestones achieved × 5.8% stake” is more likely to add to the net-cash cushion and extend runway than to single-handedly take market value to 5 times. Using it as a pillar of a 5 times thesis overstates the size of one equity realization.

  Now look at what today’s price implies. This is the core difference between SDGR and SIDU. The current $15.85 sits at the lower end of the report’s base range ($15–22). EV is about USD 890 million, while the software segment alone, valued at peer EV/Sales of 3–6x, is worth about USD 900 million–1.1 billion. In other words, the market’s current price is roughly a conservative SOTP of “software + about USD 300 million net cash,” giving almost zero credit to owned-pipeline options and the equity portfolio (including the Ajax 5.8% stake). The embedded market expectation is “downside support exists, but we do not believe those ‘free options’ will pay off.” Put differently, the report’s most optimistic bull range is $28–42, only about 1.8–2.6 times the current price; a 5 times result over ten years, to about USD 6.0 billion and a share price far beyond the bull range, already exceeds the report’s own most optimistic scenario. That means “5 times” requires not just delivery, but delivery beyond even the most optimistic sell-side scenario, further lowering the multiplied probability.

  Horizontal comparison with LUNR, rated “Watch” in the same framework, helps place SDGR. Both are unprofitable and burning cash, but their risks are opposite. LUNR is “expensive price + high speed but unproven”: price-to-sales above 39 times and negative book equity, revenue nearly tripling, adjusted EBITDA just turning positive, a typical moonshot where growth imagination leads and valuation is already overdrawn. SDGR is “cheap price + slowing but with a real floor”: about USD 200 million of real external software revenue, about USD 300 million net cash, and repeated equity monetization history. LUNR’s danger is that if growth misses, the overdrawn valuation must give back. SDGR’s danger is that it is cheap for a reason and can become a value trap if catalysts do not arrive. This also echoes its key difference from SIDU: SIDU is “expensive price + bad business,” double overdrawn; SDGR’s valuation is not overdrawn and it has a real asset floor. It also fails the 5 times test, but the price is not bad.

  Honest landing point: the conditions for a 5 times return over ten years are numerous and each has low probability, especially the hardest link of “software slowdown + starting from revenue contraction,” plus pipeline shrinkage, need for rerating, and possible dilution. After multiplication, the overall probability is low. Baillie Gifford’s premise of “paying a high price for greatness” (evidence of greatness + upside still left in the price) is only half met at SDGR: the second half, cheap price with downside floor, is met; the first half, accelerating growth engine and evidence of greatness, is contradicted by slowdown and negative growth. But precisely because the valuation is at the lower end of base range and has real support from “software + net cash,” not overdrawn expectations, its downside risk and upside risk/reward are clearly better than a “high price + unproven” name like SIDU. The conditions are unrealistic, but the price is not a trap; overall medium-weak.

  Jun 5, 2026
• QWhy has the market not recognized all this yet? Is it because the market does not understand it, looks down on it, or cannot look far enough? What will become the “narrative inflection point”?4/10

  A

  Conclusion first: SDGR is one of the rare names in this ten-question set where the cognitive-gap direction is positive. The report explicitly judges it to be a “cheap asset undervalued by segment” (stockanalysis shows 8 analysts with an average target of $20.88, about +32% implied upside, 6 Strong Buy + 2 Hold and no Sell; SOTP bulls BofA / KeyBanc see $30–52). That is exactly the kind of “hidden value the market has not recognized” Baillie Gifford wants. But the opposite side must also be stated honestly: this is a classic “value trap candidate.” It has looked cheap on SOTP for years, while the share price has fallen from $117 in 2021 to about $15.85 today (-86%). So the answer is not simply “the market cannot see greatness,” but “the market can see cheapness, but refuses to pay for cheapness without a catalyst.” This dimension is the opposite of SIDU-type names pushed up by themes and overvalued: SDGR has a real positive cognitive gap, but the path to realization depends heavily on exogenous inflection points.

  First set the direction of the cognitive gap correctly; this must be explicit. For most growth stocks, hidden value in Q10 means “the market underestimates its ceiling / compounding.” SDGR is different. Its hidden value is “segment undervaluation” in an accounting and valuation sense. The report estimates the software segment alone, with ACV of about USD 220 million, 70% gross margin, and 10–15% growth, valued at peer EV/Sales of 3–6x, is worth about USD 900 million–1.1 billion of enterprise value, while the whole company’s current EV is only about USD 890 million. In other words, at a $16 share price, buying the software leg is almost equivalent to receiving the owned-pipeline options, equity portfolio (including the Ajax 5.8% stake being acquired by Lilly), and about USD 300 million net cash “for free.” This is a clear, verifiable, segment-sum cognitive gap, not empty narrative. It is also the core reason the report gives “Hold” rather than “Sell” and anchors downside at software + cash.

  But “cheap” alone does not explain why the market has not recognized it. Through the prism of “does not understand / looks down / cannot look far,” all three apply. First, the market does not understand it: SDGR’s business model has three legs, “selling shovels (software) + prospecting for gold itself (owned pipeline) + taking equity economics (portfolio),” and the financial statements are inherently “unclean.” A good quarter may reflect deferred recognition or one-off equity monetization; a weak quarter may be timing (for example, Q1'26 drug discovery revenue rose +124%, but the company acknowledged the main causes were accelerated recognition of deferred collaboration revenue + a project termination releasing deferred revenue, not operating growth). This lumpy revenue structure is genuinely hard for sell-side models and stable valuation yardsticks. Second, the market looks down on it: with market cap only about USD 1.12–1.18 billion, down -86% from its peak, exiled into small-cap purgatory, low liquidity, and few catalysts, institutional attention and incremental capital naturally pass it by. “A cheap little broken stock” is one of the easiest things for institutional portfolios to ignore. Third, the market cannot look far enough: current pricing ($15.85, at the lower end of the report’s base range of $15–22) is roughly only the conservative SOTP of “software + net cash,” giving almost no credit to owned-pipeline options and the equity portfolio. The market values what it can see in current cash flows and refuses to prepay for options that need time and exogenous events to realize.

  This is the standard value-trap profile, and it has to land there honestly: what is missing is not value, but catalysts. The report’s Pre-mortem lists “continued value trap” as the most likely losing scenario three years out: software growth stuck around the low teens, gross margin not fully recovered, no major drug from the owned pipeline, equity monetization occurring only sporadically, and the market still refusing to price the “free options,” leaving the stock grinding between $10–18 for three years. Cheapness is never a catalyst, and every SOTP component, software valuation and equity portfolio value included, can shrink if fundamentals deteriorate. So “cheap” is relative, not an absolute margin of safety. This is the key difference between SDGR and genuinely high-quality growth names whose quality is underpriced: SDGR’s undervaluation is real, but whether it converts into returns is not under the company’s full control.

  What can become the “narrative inflection point”? The report and current facts provide a relatively clear catalyst list. Any one of these could make the market reprice those “free options”: (1) Ajax monetization closes: Lilly announced in April 2026 that it would acquire Ajax for up to USD 2.3 billion in cash (core asset AJ1-11095, an oral Type II JAK2 inhibitor co-designed with SDGR), and SDGR owns about 5.8%; according to Q1'26 results, the transaction had been announced and was pending completion at the time of the quarter, the monetization amount was undisclosed, and it was not included in 2026 financial guidance. It is a pure upside option and the most direct catalyst to turn the “equity portfolio is free” argument from paper into cash. (2) Software ACV growth returning to 15%+, the hosted transition completing, and gross margin recovering from the current 69% to high-70s; software is the foundation of the SOTP floor, so reacceleration directly raises the segment valuation anchor. (3) A large late-stage partnership for the owned pipeline (SGR-1505 and SGR-3515 have both shifted toward finding partners), or additional milestones / royalties triggered by Nimbus’s TYK2 drug launch in 2027. (4) Bunsen, the agentic AI product with summer 2026 early access, scaling and raising platform utilization and ACV. (5) A clearer path to 2028 adjusted EBITDA breakeven, allowing the market to value it as a “profitable software company” rather than a “cash-burning small-cap biotech.”

  Honest landing point: on Baillie Gifford Q10, SDGR does have a genuine positive cognitive gap. The SOTP undervaluation is verifiable, institutional coverage is thin, and exile pricing is giving little credit to options. This is a positive relative to most names. But whether the cognitive gap turns into returns depends heavily on exogenous inflection points, especially Ajax monetization and software reacceleration. It is a classic case of “cognitive gap exists, realization path unclear.” This is the opposite of SIDU’s negative cognitive gap, where themes and passive flows pushed valuation too high; SDGR is on the ignored and undervalued side. But compared with truly high-quality growth stocks, it lacks the certainty that “time is on its side and compounding will automatically show up.” The historical baggage of being cheap for years without realization means the margin of safety only truly works at a lower price or with clearer catalysts. This is the internal logic behind the report’s “Hold” rating and ideal buy zone below $14: the direction is right, but patience and catalysts are both necessary.

  Jun 5, 2026