Schrödinger: A 6.0 Million USD Q2 Profit Over a 41.5 Million Operating Loss, and a 20% Rally Driven More by EV/ACV Re-Rating From 3.8 to 4.9 Times Than by ACV Growth

Management has also described an approximate annual revenue-recognition sensitivity of $2–3 million for each percentage-point increase in hosted mix. Handle that one carefully. Applied mechanically to the 16-point year-over-year change in Q2 hosted mix, it would imply a $32–48 million quarterly revenue drag against a Q2 software line of $32.5 million in total, which is plainly invalid. The sensitivity is an annual portfolio-level bridge, not a formula for restating an individual quarter. ACV is the better like-for-like demand measure. H1 2026 ACV was about $58 million, roughly 19% above the prior-year first half; Q2 ACV was $29.6 million, up 27%, while Q2 ACV excluding contribution ACV was $22.6 million, up 23%. Full-year ACV guidance remains $218–228 million, or 10–15% growth over 2025's $198.5 million.

That also clears up the apparent contradiction around the widely quoted $29.6 million Q2 ACV figure. The $29.6 million is no mysterious hosted-only or new-business submetric. It is the company's total ACV for the second-quarter reporting period; $208 million is trailing-four-quarter ACV; and $218–228 million is the full-year 2026 guidance. Schrödinger defines ACV as contract value billed in the applicable period for contracts of one year or less or annually billed multi-year contracts; for multi-year contracts billed upfront, the total billed amount is divided across the contract term. It warns explicitly that ACV is independent of GAAP revenue and is neither annualized GAAP revenue nor a revenue forecast. The figures differ because the periods differ, and because renewals are heavily seasonal.

The commercial evidence behind ACV reads better than the headline revenue decline. In 2025 the top-20-pharma customer cohort generated $80.8 million of ACV, up 15.3%; commercial ACV rose 7.0% to $177.4 million; average ACV among commercial customers spending more than $1 million increased to $3.9 million from $3.3 million. Gross dollar retention held at 96%. Against that, net dollar retention fell to 100% from 113%, and the number of $1-million-plus commercial customers fell to 27 from 29, although two former customers were acquired. The evidence supports deeper use inside major accounts. It does not support an across-the-board SaaS-style land-and-expand acceleration.

One prior load-bearing claim improves under verification. The latest 2025 annual disclosure says all 20 of the world's top 20 pharmaceutical companies, ranked by 2024 revenue, licensed Schrödinger solutions. The earlier “19 of 20” claim is no longer the current figure.

The other prior claim, roughly $700 million of “cumulative realizations across Nimbus, Morphic and Ajax,” needs more correction. Primary disclosures do not support treating $700 million as equity realizations from those three companies. Disclosed cash generated by equity distributions and sales is much smaller. Schrödinger's own broader language refers to cash or proceeds from drug-discovery, partnership and business-development activity. That bucket includes upfront collaboration cash such as $55 million from Bristol Myers Squibb in 2020 and $150 million from Novartis in 2025, milestones, equity dispositions and other program economics. By September 2026 the company's stated aggregate exceeded $750 million. The record is a good one, but the label matters: recurring collaboration economics and the sale of an appreciated equity stake have very different valuation characteristics.

Two developments since the prior report improve the strategic picture. Bunsen, Schrödinger's agentic AI “co-scientist,” entered early access in July; a full commercial version was expected by year-end 2026. As of the research date there is no disclosed standalone price, no Bunsen ACV and no separately reported revenue. Treating it as an established growth business would be premature.

The Bristol Myers Squibb agreement has more substance to it. Announced August 5, it is a strategic software agreement to deploy Schrödinger's platform and Bunsen across the BMS research organization, with the companies also working together on new Bunsen capabilities, large-scale chemical exploration and RetroSynth. No contract duration, ACV or financial consideration was disclosed. Do not conflate it with the companies' 2020 drug-discovery collaboration, which carried a $55 million upfront payment and potentially $2.7 billion of milestones. The 2026 transaction is, on the disclosed facts, an enterprise software deployment; the existing discovery relationship is separate.

Tectora adds another piece to the portfolio model. On September 9 Schrödinger announced that it had co-founded the immunology and inflammation company with NEA and RA Capital. Tectora raised $55 million in Series A capital; Schrödinger contributed two early-stage programs, SDGR-4594 and SDGR-8139, in exchange for equity plus future milestones and royalties. The percentage ownership and carrying value were not disclosed. The transaction cuts the direct capital needed for those assets while preserving upside, the same model that made Nimbus, Morphic and Ajax financially valuable.

The stock has risen from the previous house anchor of $15.85 on June 5 to $19.02 on September 11, a 20.0% move. The business did improve over that interval: Ajax closed and converted a stake into cash, Q2 ACV was strong, BMS validated large-scale enterprise deployment, Bunsen reached early access, and Tectora supplied another example of external financing for Schrödinger-originated assets. A rough enterprise-value bridge, though, says the share-price move owed more to re-rating than to the increase in software ACV alone. Using the prior $15.85 share price, roughly 74 million shares, Q1 liquidity near $406 million and trailing ACV of about $201 million gives an approximate EV/ACV multiple around 3.8 times. At $19.02, using the $1.44 billion market capitalization, June 30 liquid resources of $418.8 million and trailing ACV of $208 million, the multiple is about 4.9 times. That is roughly 28% multiple expansion against only about 3.5% sequential growth in trailing ACV. The calculation is necessarily approximate, since the cash dates differ and it does not mark private investments. It establishes the direction.

The refreshed portrait is a company in transition: recurring software economics are getting cleaner while accounting revenue temporarily gets uglier, and management is increasingly externalizing the capital burden of its drug pipeline.

That is a materially sturdier story than one built on “AI drug discovery.” Schrödinger's core technical proposition starts with molecular mechanics and statistical thermodynamics: FEP estimates relative free-energy differences by simulating molecular systems and sampling changes between related ligands. Machine-learning drug-discovery platforms generally infer useful relationships from training data, high-dimensional biological data or learned molecular representations. Schrödinger increasingly uses AI itself, but FEP+'s differentiator remains its physics-based ground truth. The company markets FEP+ as having roughly 1 kcal/mol predictive accuracy in broad validation studies. Commercial adoption across the top 20 pharma companies is strong evidence of utility. What public evidence does not establish is a controlled, industry-wide causal claim that Schrödinger customers have higher clinical success rates than users of competing approaches.

For valuation, the dual identity makes conventional P/E almost useless. At $19.02, the market capitalization is $1.44 billion. June liquidity was $418.8 million and there was no material traditional funded debt, leaving a rough enterprise value of $1.02 billion before lease liabilities and before recognizing investment stakes. Against trailing ACV of $208 million, that is 4.9 times ACV; against the midpoint of 2026 ACV guidance, about 4.6 times. That already resembles the valuation territory of specialist software peers. So the stock does not look “cheap because it is 83% below the 2021 high.” It looks approximately priced for a decent software franchise, with relatively little explicit value left over for pipeline optionality once future cash burn is reserved.

That 83% figure deserves perspective of its own. The all-time closing high was $113.09 on March 1, 2021; the current $19.02 is 83.2% below it. The 2021 price was set in an extraordinarily permissive market for long-duration biotech and technology assets. Distance from that high says nothing about today's fair value.

Vertical history, financial evolution and capital-market narrative

Schrödinger began as a scientific-software company long before “AI drug discovery” became a public-market category. Richard Friesner and William Goddard founded it in 1990, carrying academic computational-chemistry backgrounds into a commercial effort to make molecular modeling usable by working scientists. The first software sale came in 1992, to Brookhaven National Laboratory. That origin matters: the drug-development portfolio was layered onto an existing tools franchise rather than being the reason the company existed in the first place.

Its history divides naturally into five stages.

The first was the long software-validation period from 1990 through roughly the late 2000s. The product was scientific infrastructure: force fields, molecular mechanics, docking, quantum chemistry, visualization and eventually more sophisticated free-energy calculations. Customers bought better computational estimates so that less chemistry had to be synthesized and tested experimentally. The moat accumulated slowly, through code, force-field calibration, workflow integration and users trained on the software. Nothing in the public market pushed the company to turn every scientific advance into a new narrative.

The second stage was the emergence of a stronger predictive platform and the decision to participate directly in molecule economics. FEP+ and related workflows let Schrödinger turn increasingly powerful compute into compound rankings accurate enough to be useful prospectively during lead optimization. Rather than limit itself to licensing tools, the company began applying them to collaborations and co-founded ventures. Nimbus became the canonical example: Schrödinger helped build a computationally driven discovery company, retained economics and later received cash when assets were sold. Relay, Morphic, Ajax, Structure and other ventures extended the template. The logic was simple. If the software materially improved discovery, licensing it for a few million dollars captured only a small part of the economic value. Equity, milestones and royalties let Schrödinger retain a claim on successful molecules.

The third stage was the 2020 IPO and market re-rating. Schrödinger priced its IPO at $17 on February 5, 2020, above the indicated range, offering roughly 11.9 million shares for about $202 million of gross proceeds before the overallotment option; trading began February 6. Full exercise of the underwriters' option took the total sale to roughly 13.6 million shares and gross proceeds to about $232 million. On the post-IPO share count, that implied an equity value of roughly $1.1 billion.

The public-market story quickly became much larger than software. In 2020 revenue was $108.1 million, software revenue $92.5 million, software gross margin 81%, and total ACV $92.1 million. At the same time it was expanding its internal pipeline, sitting on appreciating stakes in companies such as Relay and Morphic, and had just signed a Bristol Myers Squibb discovery agreement with $55 million upfront and up to $2.7 billion of potential milestones. In a zero-rate market willing to capitalize distant biotech and platform optionality at aggressive multiples, the shares eventually closed at $113.09 on March 1, 2021, more than six times the IPO price.

The fourth stage, from 2022 through 2024, tested what the market had paid for. Revenue grew, especially software, but research spending rose much faster. Total revenue increased from $181.0 million in 2022 to $216.7 million in 2023 and then dipped to $207.5 million in 2024. Software revenue continued climbing from $135.6 million in 2022 to $159.1 million in 2023 and $180.4 million in 2024. Operating losses still widened, from $146.8 million in 2022 to $177.4 million in 2023 and $209.3 million in 2024, as the company funded platform R&D and an expanding proprietary clinical pipeline.

The 2023 income statement is the cleanest historical illustration of the accounting problem investors face today. Schrödinger reported $40.7 million of net income on a $177.4 million operating loss, because other income came to $220.4 million, including a $147.2 million gain/distribution from the Nimbus transaction and $53.5 million of fair-value gains on equity investments. The business had not suddenly turned operationally profitable. Portfolio monetization had swamped the operating loss for one year.

Higher interest rates and a much colder biotech-capital market simultaneously reduced the present value of distant pipeline cash flows. The market stopped valuing Schrödinger like a scarce, almost unlimited “computational drug discovery” option and started asking what the software franchise could earn and how much capital the proprietary pipeline would eat. That re-rating was economically rational even though the technology had not become obsolete.

The current stage began in 2025, with three changes running at once: software contracts shifting toward hosted and ratable recognition, consolidated operating costs coming down, and more drug programs moving toward partnership or external financing instead of carrying all clinical-development costs in house. In 2025 revenue rose 23.3% to $255.9 million, software revenue increased 10.6% to $199.5 million, drug-discovery revenue rose to $56.4 million, operating expenses fell 9.3% to $309.5 million, and the net loss narrowed to $103.3 million from $187.1 million. Management set a target of positive adjusted EBITDA by the end of 2028.

The $150 million upfront payment from Novartis, received in January 2025 once the collaboration cleared antitrust review, was especially important to cash flow. It helped push 2025 operating cash flow positive at $13.9 million even though the company lost $103.3 million. Accounts receivable also fell by $152.7 million, while deferred revenue declined $29.0 million. So 2025 operating cash flow should not be treated as proof that the operating model had reached self-funding.

The six-year financial record shows a real software-growth franchise embedded inside a capital-consuming consolidated company.

† Q2 2026 software gross margin was 71%; the company did not present a directly comparable H1 software-margin figure in the headline release. Sources: company annual and quarterly disclosures.

From 2020 to 2025 total revenue compounded at about 18.8% annually and software revenue at about 16.6%. The growth is real. The return on the capital used to generate it is not yet attractive at the consolidated level, because much of the gross profit has been recycled into R&D. Traditional ROIC and ROE come out negative, or badly distorted by equity-investment gains. That is a feature of the current business model, not a temporary accounting quirk.

The balance sheet has been the main protection against that investment intensity. At June 30, 2026 Schrödinger held $287.9 million of cash and cash equivalents, $5.1 million of restricted cash and $125.9 million of marketable securities, totaling $418.8 million. It also reported $36.5 million of equity investments. Traditional financial debt is not the principal solvency issue here; operating lease obligations were much more significant than borrowings. The question is cash runway and economic dilution, not refinancing stress.

Stock-based compensation is material. It ran $43.0 million in 2025, equivalent to roughly 17% of revenue, and the common share count rose from 63.7 million at the end of 2024 to 64.5 million at the end of 2025, on top of 9.16 million limited-common shares. By July 29, 2026 about 74.8 million common and limited-common shares were outstanding in aggregate. Part of the reason the cash balance is strong is that shareholders have absorbed meaningful equity compensation over time.

Management has proven that it can monetize scientific assets; it has not yet proven that the consolidated company can compound per-share cash earnings without periodically relying on collaboration cash, equity realizations or dilution.

Even so, the equity-monetization record is too long to write off as luck. The main disclosed realizations separate from the broader business-development proceeds as follows.

Primary disclosures show the individual events and cash flows; the company's cash-flow statements report $48.8 million of equity-sale proceeds in 2024, $17.7 million in 2025 and $67.0 million in H1 2026.

Adding the readily identifiable equity distributions and disposition proceeds produces a figure in the mid-$300 millions, not $700–750 million. The larger company figure includes business-development proceeds such as collaboration upfronts and milestones. That distinction changes how the history should be read. Schrödinger has repeatedly turned computational-discovery work into cash through several channels, and that is a capability. The cash is still episodic and asset-specific. Putting a multiple on cumulative realized proceeds would capitalize past exits twice; value the remaining stakes and contractual rights prospectively instead.

The portfolio remaining at the end of 2025 included interests in Nimbus, Bright Angel, Apollo and Lakshmi vehicles and Structure, among others; Ajax has since been acquired. At year-end 2025 Schrödinger disclosed a 1.1% Nimbus interest, 31.5% of Bright Angel, 7.9% of Apollo LLC, 5.3% of Lakshmi LLC and 1.6% of Structure, although the economic meaning differs because some vehicles primarily hold rights to future transaction payments. Private stakes are not reliably markable at current market value from public information.

The stock-price history mirrors these narrative shifts.

After the $17 IPO, the market took the shares to the $113.09 March 2021 closing peak as zero rates, biotech enthusiasm and the platform narrative converged. That valuation then unwound as internal R&D expenditure rose and long-duration growth multiples collapsed. The stock has never come near its 2021 peak since.

The recent event tape teaches more than the old peak does. Lilly's 2024 acquisition of Morphic crystallized a valuable stake; the late-2024/early-2025 Novartis collaboration brought a $150 million upfront payment; Q1 2026 produced a roughly 6% post-results selloff even though revenue beat estimates, as investors focused on the wider loss and hosted-revenue transition; Q2 then delivered faster ACV growth, the Ajax monetization, the BMS enterprise agreement and Bunsen early access. Tectora followed in September.

So the June-to-September advance has a factual basis, but the EV/ACV bridge shows investors have also raised the multiple they will pay for those facts. The current debate has shifted from “will the software franchise keep growing?” toward “how much of the apparent revenue weakness is accounting timing, and what value should be assigned to the drug-discovery engine without pretending its realizations are recurring earnings?”

Business model, moat, industry and competitors

Schrödinger's business machine produces four economically different outputs: commercial software ACV, GAAP software revenue, drug-discovery collaboration revenue, and investment/milestone/royalty value. Blending them into a single growth rate destroys information.

Software ACV tracks contracting activity most closely. Full-year 2025 ACV was $198.5 million, and management guides 2026 to $218–228 million. GAAP software revenue was $199.5 million in 2025 but is falling for now as hosted contracts replace upfront-recognized on-premise agreements. Drug-discovery revenue was $56.4 million in 2025 and is guided to $65–75 million in 2026 after the Ajax milestone raised guidance by $10 million. Contribution revenue was separated in the 2026 presentation and added $3.4 million in Q2.

Software is the economic anchor, with gross margins historically in the mid-70s to low-80s. The 2025 software gross margin was 74%; Q2 2026 was 71%, carrying hosted-delivery and product-investment costs through the transition. Management's 2028 objective is to return software gross margin to the high 70s as the hosted model scales.

Drug-discovery gross margins are structurally harder to interpret. Research funding and milestones can carry high incremental margins when the underlying discovery work has already been expensed through R&D, while active research programs can create sizable cost of revenue. Quarterly margin therefore moves with contract mix rather than with productivity alone. That is why a $10 million milestone cannot be annualized into $40 million of “run-rate” earnings.

The fixed-cost base is mainly scientists, software engineers, clinical teams, sales personnel, computing infrastructure and corporate functions. Physical capital expenditure is small: purchases of property and equipment were $13.4 million in 2023, $7.3 million in 2024 and only $1.4 million in 2025. Most economic reinvestment is expensed immediately through R&D rather than capitalized on the balance sheet.

A cost base shaped like that converts growth into margin if software ACV can grow 10–15% while opex stays roughly flat. It also explains why the 2028 adjusted-EBITDA target is feasible without extraordinary revenue growth. The reverse is what makes an ACV slowdown dangerous: the company cannot cut platform R&D indiscriminately without damaging the technology that supports both software and discovery.

The first genuine moat is workflow and scientific validation. FEP+ is more than a generative model proposing molecules. It uses molecular mechanics, force fields and statistical sampling to estimate changes in binding free energy between related compounds. In a lead-optimization loop, that can let chemists rank compounds before synthesis and concentrate laboratory work on candidates predicted to have better potency, selectivity or other properties. Schrödinger combines those simulations with AI and machine learning rather than treating the two methods as mutually exclusive.

The company's “gold standard” language should be treated as marketing rather than a formal industry designation. Adoption is the better evidence. All of the top 20 global pharmaceutical companies by 2024 revenue were software customers in 2025, gross commercial dollar retention was 96%, and large-account ACV continued to rise. These customers have sophisticated internal computational chemistry groups and many alternative tools. Repeated renewal at high spend levels is more probative than a slogan.

The second moat is cumulative scientific know-how. Force fields, workflow design, error detection, system preparation and sampling protocols embody decades of experimental feedback. FEP+ depends on more than raw compute. A competitor can buy GPUs; reproducing chemistry-specific validation across a wide chemical space is a slower process. That moat is real, though it can erode as open-source methods, machine-learned potentials and better general-purpose models improve.

The third moat is organizational integration. LiveDesign, modeling tools and now agentic systems such as Bunsen make Schrödinger part of a research workflow rather than a single calculation purchased in isolation. Large deployments raise switching costs, because researchers build habits, shared data structures and internal processes around the platform. That is what makes the BMS agreement strategically important. It says a sophisticated pharmaceutical customer is prepared to deploy the stack broadly, even though the disclosed economics are insufficient to quantify its valuation impact.

The fourth, less conventional moat is what the company has learned about converting simulation into economic ownership. Nimbus, Morphic, Ajax and now Tectora show a repeated ability to identify assets, supply computational discovery and structure arrangements that leave Schrödinger with equity or downstream economics. That capability cannot be valued like software recurring revenue. It is also more than a marketing moat, because several independent transactions have paid cash.

The weak point is profitability. Competitors do not need to replace FEP+ technically to pressure its valuation. They need only persuade customers that acceptable simulation is available cheaper, bundle modeling with broader development software, or offer AI workflows that cut the dependence on expensive high-accuracy calculations across enough use cases. If that happens, the moat could remain scientifically respectable while becoming financially less valuable.

Industry structure is fragmented because “computational drug discovery” bundles several businesses that customers purchase for different reasons. At one end are scientific-software vendors. At another are data- and AI-native drug developers that retain molecules themselves. Between them are contract discovery, model-informed drug-development services, automated laboratories and platform collaborations. A single market-size estimate is more likely to obscure Schrödinger's opportunity than to illuminate it.

The industry's profit pool also shifts by stage. Proven enterprise software can generate high gross margins and repeat purchases. A preclinical asset has negative cash flow but enormous option value if it later becomes a differentiated clinical drug. The owner of a successful approved molecule captures far more value than the software provider that helped design it. Schrödinger deliberately spans those layers to avoid leaving all of the asset economics with customers.

Its cycle exposure is correspondingly mixed. Pharmaceutical R&D budgets are relatively defensive, helping software. Small-biotech funding and M&A activity are cyclical and rate-sensitive, affecting collaboration formation and the value of equity stakes. Proprietary pipelines are governed by scientific and regulatory cycles. Public-market valuation remains highly sensitive to real rates because much of the pipeline value lies years in the future.

Regulation is not a direct license barrier for the software franchise, but it matters through customers. FDA and other regulators ultimately determine whether proprietary and partnered candidates can generate milestones and royalties. Cybersecurity, data residency and validation requirements also matter as hosted deployments increase. The move toward hosted software can improve central support and recurring recognition but also shifts more responsibility for availability, security and customer data to Schrödinger.

Horizontally, Schrödinger has ample competitors, but there is no single perfect listed comparable.

Certara is the closest public-company benchmark for monetizing scientific software and expertise without owning a similarly large proprietary drug pipeline. Its center of gravity is model-informed drug development, biosimulation, clinical pharmacology and regulatory workflows rather than atomistic lead optimization. Certara says its customer base exceeds 2,600 biopharma companies, academic institutions and regulators, and that its customers were involved in more than 90% of FDA novel drug approvals between 2014 and 2025. Q2 2026 revenue was $93.3 million, up about 1%, with software revenue of $48.8 million, up 4%. Its services component is more mature and its growth weaker, but it already generates cash, which makes it a useful valuation anchor for the software half of Schrödinger.

Simulations Plus is smaller and more specialized. It sells modeling and simulation software alongside consulting services in drug development. Q3 fiscal 2026 revenue was $21.9 million, up 7%, while software revenue was $12.6 million and roughly flat. In June 2026 it agreed to be acquired by Altaris at $18.50 a share, implying an equity value around $375 million. That deal matters here because it supplies a real-money benchmark for a scientific-software company instead of a volatile public-market quote.

Recursion sits on the opposite side of the comparison. It is essentially a pipeline/platform company using AI, automated experimentation and large biological datasets to identify and develop drugs. In Q2 2026 it generated only $7.7 million of revenue, lost $131 million and used $105.9 million of operating cash, while holding $556.8 million of cash and restricted cash. Its roughly $1.7 billion market capitalization is predominantly a valuation of platform and pipeline option value, not of recurring software earnings. That makes Recursion useful for understanding what investors pay for drug-discovery optionality, but a poor multiple for Schrödinger's licensing business.

XtalPi is technologically closer in some respects, combining physics, AI and automated laboratories, and it has also moved toward pipeline licensing. Its H1 2026 disclosures carry the same analytical hazard as Schrödinger's: reported revenue can be dominated by a large licensing payment, while underlying service and AI-for-science activity follows a different growth path. The comparison reinforces the need to separate platform demand from milestone accounting.

AbCellera and Absci are often placed in the same “AI-enabled drug discovery” basket, but both sit farther from Schrödinger's economic model. Their technology centers more on biological discovery and protein or antibody design, and their valuations depend heavily on pipeline and partnership outcomes. They work as sentiment comparables, not as appropriate anchors for FEP+ software.

The current market values these businesses very differently.

† SDGR Q2 profit was dominated by Ajax/equity economics and should not be treated as recurring. ‡ Certara also carried roughly $295 million of term borrowings around Q1 2026. Sources: company filings and market data.

The business reasons behind the valuation gaps matter more than the table. Certara earns a lower growth multiple because its revenue growth is slow and its services mix dilutes software economics, but investors can see a path to conventional EBITDA. Simulations Plus's take-private valuation supplies a roughly mid-single-digit revenue-multiple ceiling for a small scientific-software asset without Schrödinger's pipeline. Recursion trades almost entirely on future biological and clinical value. Schrödinger sits between them. That is why a single peer multiple produces nonsense.

Schrödinger's niche is a premium scientific-software supplier that has successfully retained a slice of the drug assets its own technology helps create. Its most defensible profit pool is enterprise licensing. Its largest potential upside comes from economics that cannot be forecast with comparable precision.

Current fundamentals and valuation

The last four quarters show why headline revenue is especially misleading now.

† Q3 2025 ACV was $38.3 million including contribution ACV; annual renewal concentration makes quarterly comparisons highly seasonal. Q1 and Q2 2026 figures are the period ACVs, not annualized revenue.

Q4 2025 and Q2 2026 both showed GAAP profits or near-profits because investment gains outran operating losses. In Q4 2025 the change in fair value of equity investments was roughly $47.0 million; in Q2 2026 fair-value gains were $45.9 million. That these gains keep recurring is evidence the portfolio has economic value. Their quarterly timing makes net income a poor operating KPI.

The cleanest current operating indicators are ACV, retention, gross margin and opex discipline. Q2 ACV growth of 27% ran far ahead of reported software revenue growth of negative 10%. Full-year ACV guidance of $218–228 million remains consistent with 10–15% underlying contracting growth. Operating expenses in Q2 fell 6% to $74.0 million, and management still expects full-year opex below 2025. That combination has to persist for the 2028 adjusted-EBITDA target to become credible.

The H1 numbers show the hosted transition more starkly. H1 2026 GAAP software revenue was $68.1 million, down 23.7% from $89.3 million a year earlier. First-half ACV was roughly $58 million, up about 19%. The two lines measure different things: one reflects revenue-recognition timing, the other is intended to normalize contract duration, billing and deployment models. A precise “GAAP revenue as though hosted adoption had not changed” cannot be calculated from public data, because Schrödinger does not disclose the contract-level recognition schedules required to build that counterfactual. That leaves ACV as the defensible like-for-like measure.

None of this is a free pass for management. ACV itself has to be corroborated by customer behavior. The 2025 net-dollar-retention decline to 100%, versus 113% in 2024, argues against reading every point of ACV growth as accelerating product demand. Large pharma accounts are expanding. The broader commercial cohort is more mixed.

Keep Bunsen out of near-term forecasts until management reports a price, separate bookings or incremental ACV. Its early-access launch and BMS deployment establish that the product exists and is being used by a sophisticated customer. They do not establish a new revenue curve. Full commercial launch was expected by year-end 2026.

The owned clinical portfolio has also become less capital-intensive by design. Schrödinger has said it is exploring partnerships for SGR-1505, its MALT1 inhibitor, and SGR-3515, its Wee1/Myt1 inhibitor rather than committing to fund broad later-stage development itself, and had been working to complete Phase 1 packages for both. That lowers the cash requirement if deals are signed. It also means shareholders may receive only a negotiated slice of downstream economics rather than the full product value.

Valuation starts with the cash-flow passthrough test. Conventional P/E fails immediately: trailing GAAP earnings are negative, and even the profitable quarters contain large investment gains. Owner earnings are negative on a normalized basis too.

2025 positive cash flow was heavily affected by collaboration cash and working-capital movements, including the $150 million Novartis upfront received in January.

A five-year OCF-to-net-income “quality ratio” carries no economic meaning for Schrödinger, because numerator and denominator change sign for different reasons: operating cash flow moves with large upfront collaboration payments, net income with mark-to-market and equity-disposition gains. The annual observations above show the problem. For a mature industrial company the ratio would be a useful earnings-quality test. Here what it tells you is that normalized owner earnings remain negative.

Maintenance capex versus growth capex is not separately disclosed. Physical capex is small enough that the distinction is not the valuation bottleneck anyway. The economically important maintenance expenditure is scientific R&D and software development, much of which is expensed. Treating all R&D as discretionary “growth capex” would overstate owner earnings. A sensible valuation, then, does not capitalize a positive FCF stream that does not yet exist.

Headline P/E tells you nothing here, and neither does owner-earnings P/E or FCF yield, because normalized owner earnings are below zero. That is why the SOTP below uses ACV for software and rNPV for drug assets.

Set against its own history, today's valuation sits far below the 2021 speculative extreme. A precise percentile would be false precision, since the company has changed its revenue-recognition model, ACV disclosure and portfolio composition since listing. EV/ACV is the more useful current statistic. At $19.02, equity value is about $1.44 billion. Subtracting the June liquid-resources balance of $418.8 million gives about $1.02 billion of rough enterprise value before lease liabilities and investment assets. That is 4.9 times trailing ACV of $208 million and 4.6 times the midpoint of 2026 guidance.

The pure-software peer anchors put that multiple in context. Certara's current enterprise value is roughly in the mid-$1 billion range after accounting for cash and term debt, against annualized Q2 revenue in the upper-$300 millions, around the mid-3-times revenue area. Simulations Plus's $18.50 take-private price implies roughly $375 million of equity value against a recent quarterly revenue run-rate around $88 million annualized, or a little above four times revenue before adjusting for net cash. Those are not apples-to-apples ACV multiples, but they argue against valuing Schrödinger software at the double-digit sales multiples common in 2021.

The SOTP uses those anchors but gives Schrödinger a range reflecting stronger ACV growth and its specialized FEP position. It then values the drug side separately.

The pipeline rNPV is deliberately conservative. SGR-1505 and SGR-3515 are early clinical assets whose ultimate indications, partner economics, probability of approval and commercial peak sales all remain uncertain. Rather than pretend to know those values precisely, I assign aggregate risk-adjusted values consistent with early clinical-stage optionality and raise them only in the optimistic case. Collaboration rights are separated from wholly owned assets so the equity monetization engine is not double-counted.

The software multiple is the main valuation judgment in the table. Four times ACV roughly matches specialist-software transaction/public comps, with no credit for heroic margin expansion. Five times assumes 10–15% ACV growth, high-70s eventual software gross margins and a credible 2028 EBITDA bridge. Six times needs evidence that Bunsen, broad enterprise deployments and higher throughput extend growth without costing pricing power. The forward-burn reserve is there to stop the valuation crediting every dollar of today's cash as excess capital.

At today's market value, the pipeline is being valued at roughly zero, not at a large positive number.

One way to see it is to value software at 4.5–5.0 times ACV. At 4.5 times the 2026 midpoint, software is worth about $1.00 billion. Add roughly $350 million of usable cash after a burn reserve and about $50 million for stakes, and the total reaches around $1.40 billion before any meaningful value for the owned pipeline or collaboration rights, almost exactly today's market capitalization. At five times ACV the same exercise already exceeds the market capitalization before pipeline value. The answer is sensitive to the multiple and the burn reserve. The central point survives either way: shareholders at $19 are not paying an obvious hundreds-of-millions premium for clinical success.

The valuation scenarios then translate into investment-return ranges.

This is valuation-scenario analysis within a research framework, not investment advice.

The expectation gap here is an odd one. Bears looking only at GAAP software revenue can conclude the commercial business is shrinking, which understates demand because of the hosted conversion. Bulls looking only at ACV can conclude the transition is purely cosmetic, which overstates the evidence: NDR is only 100%, gross margin is temporarily lower, and the company has not proved the hosted model will accelerate customer expansion.

The next earnings print will be judged mainly on Q3 ACV. Management guided to $41–45 million excluding contribution ACV, against $38.3 million total ACV in Q3 2025, which included $2.2 million of contribution. Full-year guidance then implies another very large Q4 renewal period. A Q3 result at or above the high end with unchanged full-year guidance would validate demand; a miss below $41 million would make the accounting-transition explanation less convincing.

The second expectation gap is BMS. Reading a large-scale BMS deployment as commercial validation is reasonable. Building a revenue estimate from it is not, because neither ACV nor duration has been disclosed. The first explicit evidence that Bunsen lifts ACV, pricing or retention would be more valuable than another product announcement.

The third is the proprietary pipeline. The market currently assigns little explicit value to it under a reasonable software valuation. That creates upside if partnership deals validate SGR-1505 or SGR-3515, but it does not eliminate loss risk because those assets are also part of the justification for ongoing R&D.

The margin-of-safety check is less favorable than the base-case upside suggests. The conservative SOTP is about $18.7 a share, slightly below the current $19.02. The shares trade at a small premium to the conservative case. Margin of safety to that case is zero.

The most fragile base-case assumption is the five-times ACV software multiple. Cut it to 70%, or 3.5 times, and leave every other base assumption unchanged, and value drops by roughly $335 million, or about $4.5 a share. Base value falls from about $24.3 to roughly $19.8, almost exactly the current price. That is a warning against treating the 28% base upside as a high-confidence spread.

If owner earnings remain flat for three years at today's normalized negative or approximately zero level, there is no operating cash yield to shareholders. Even without needing an exact Treasury quote, a non-positive owner-earnings yield is below any positive 10-year government-bond yield. Under the framework specified in the brief, there is no margin of safety at this buy price.

Margin-of-safety sufficiency verdict: none.

Risks, catalysts and tracking dashboard

The first permanent-loss risk is that ACV growth is weaker than the hosted-accounting explanation suggests. I assign medium probability and high impact. Watch ACV, NDR and large-customer ACV. If full-year ACV falls below the $218 million floor, or commercial NDR drops below the current 100% while hosted penetration keeps rising, the transmission path is direct: the market stops treating reported revenue weakness as timing, cuts its long-term growth estimate, and the software multiple can slide from roughly five times ACV toward three or four.

The second is persistent cash consumption. Probability is medium; impact is high over three to five years. H1 2026 operating cash use was $48.9 million even with Ajax proceeds landing in investing cash flow, and accumulated deficit reached roughly $683 million by June. The current balance sheet buys several years of flexibility, but $80–120 million of normalized annual cash burn would eat a meaningful portion of current liquidity before 2028. The indicators to follow are operating cash flow, opex and cash plus marketable securities. A fall below $300 million of liquid resources without clear EBITDA convergence would raise the probability of equity issuance, or force deeper pipeline cuts.

The third is pipeline-option failure. Probability is high for any individual early-stage program and medium at the portfolio level; impact is medium to high. Schrödinger's model spreads this risk across owned, partnered and co-founded assets by design, but SGR-1505 and SGR-3515 still need clinically persuasive data and attractive partners. Failure to partner them would either cut pipeline rNPV or force management to choose between spending more directly and abandoning value. The test is whether one or both receive credible partnership terms after Phase 1 packages are complete.

The fourth is technology commoditization. Probability is medium over five years and impact is high. The threat is unlikely to be a single FEP clone. It is a stack in which machine-learned potentials, internal pharma models, cloud compute and competing scientific software make accurate enough predictions cheaper or faster across a large share of design work. Certara is already integrating AI agents into its biosimulation stack, while XtalPi is investing heavily in AI-for-science and automated laboratories. The observable evidence would be falling retention, pricing concessions, slower ACV-per-large-customer growth or customer migration toward bundled alternatives.

The fifth is dilution. Probability is medium; impact is medium. Stock-based compensation was $43 million in 2025, and the share count keeps rising. A company can create enterprise value while per-share value stagnates, if compensation and future financing absorb too much of the gain. The measure here is fully diluted share growth against ACV growth; sustained dilution above roughly 3% annually would deserve a larger haircut to long-duration per-share scenarios.

The sixth is valuation compression. Probability is medium, impact high, and the reason is that current owner earnings are negative. A five-times ACV valuation rests on confidence that software growth and future margins will converge. If long-duration biotech and software multiples de-rate, or the 2028 EBITDA target slips, there is no earnings yield to cushion the fall.

Positive catalysts over the next year are unusually concrete. Q3 ACV at or above $45 million would strengthen the demand argument. Bunsen moving from early access to a priced commercial product by year-end, followed by disclosed customer uptake, would turn an AI narrative into a measurable one. A partnership for SGR-1505 or SGR-3515 could push clinical spending outside the company while crystallizing rNPV. Further Structure, Tectora, Nimbus or collaboration milestones would add to the portfolio-monetization record.

The main negative catalysts are the mirror image: ACV below guidance, a further decline in NDR, software gross margin staying near 70% rather than recovering, renewed opex growth, failure to place the lead clinical programs with partners, or an extended period in which portfolio proceeds dry up while operating losses continue.

The November 4 earnings date is a third-party estimate, not a company-announced date; both MarketBeat and Zacks currently estimate November 4, while another calendar service gives a wider early-to-mid-November window. Treat it as provisional.

ACV and retention belong at the top of the dashboard: they are what separates accounting transition from genuine weakening. Software gross margin tests whether hosted delivery eventually produces the economic model management has promised. Cash and opex decide whether the company reaches 2028 without sacrificing per-share value. Bunsen matters only once its commercial contribution can be measured.

Cross-synthesis, rating, sources and uncertainties

Vertically, Schrödinger has proven something relatively few “platform biotech” companies have: pharmaceutical customers will repeatedly pay it for software over decades, and independent buyers have repeatedly paid real cash for drug assets or companies in which Schrödinger retained an economic interest. Those are two distinct forms of external validation. One says the platform works as a tool. The other says at least some molecules generated with its help became valuable enough for third parties to acquire or fund.

The history also shows the limitation: the software franchise has not yet funded the whole organization. From 2022 through 2025, rising scientific and clinical investment pushed consolidated operating losses far beyond the gross profit software generated. Years that looked profitable, such as 2023, were profitable because portfolio gains and distributions swamped the operating loss. Q2 2026 is another version of the same thing.

So past success came from a mixture of technology advantage, management's willingness to retain asset economics, access to external capital and favorable timing in individual drug transactions. Calling it luck would ignore the repeated Nimbus, Morphic, Ajax, Structure and Tectora pattern. Calling it a recurring earnings machine would ignore how sparse, path-dependent and asset-specific those outcomes are.

The current strategy lines up with shareholder economics better than the 2022–24 model did. The company is cutting opex, asking ACV rather than GAAP software revenue to carry the job of measuring demand through the hosted transition, and looking for partners or external vehicles to absorb more of the later-stage drug-development cost. Tectora is a particularly clean example: two Schrödinger-originated programs move into a separately financed company backed by specialist venture investors while Schrödinger keeps equity, milestones and royalties. That is capital recycling rather than capital abandonment.

Horizontally, the software moat remains stronger than the biotech moat. Certara has broader penetration downstream in model-informed development and regulatory work; Simulations Plus offers a narrower simulation-and-services model; Recursion and XtalPi pursue more AI- and data-intensive discovery architectures. Schrödinger's differentiation is its depth in atomistic physics, FEP and integrated medicinal-chemistry workflows. Customers do not have to believe in a futuristic autonomous drug lab to justify buying the software. They need the software to help make better molecule-level decisions today.

That distinction earns its keep in a crowded AI narrative. Bunsen makes Schrödinger more competitive at the interface layer, but it does not change the fundamental valuation thesis until it shows up in ACV, retention or pricing. The BMS deployment carries more weight than Bunsen's launch announcement, because a sophisticated customer is actually putting the software into large-scale use. The missing economics are what stop it from being capitalized aggressively today.

The market's likely misjudgment is two-sided. Investors focused on GAAP software revenue can underestimate the business because hosted conversion moves revenue from upfront to ratable recognition. Investors focused on “AI drug discovery” and historical exits can overestimate it by turning episodic milestone and equity proceeds into a pseudo-recurring profit stream.

The ACV evidence currently favors the company on the first dispute. Q2 ACV grew 27%, full-year guidance still calls for 10–15% growth, top-pharma spend is increasing and gross retention remains high. The earnings-quality evidence favors the bears on the second. Removing Ajax and equity marks leaves a deeply loss-making operating structure.

The result is a subtle valuation. At $19.02, investors are not paying 2021-style prices for pipeline optionality. A reasonable 4.5–5-times ACV value for software plus a forward-adjusted cash balance accounts for most or all of the market capitalization, which makes drug success genuinely additive. The stock is still not priced below conservative intrinsic value. The margin-of-safety test fails because a lower software multiple can erase the entire base-case upside.

This is why the rise since June is only partly a fundamental revaluation. The Ajax closing improved liquidity and validated another co-founded asset, the BMS deployment strengthened the enterprise-software story, Tectora cut the direct funding burden of two early programs, and ACV execution improved. Those developments raise fair value. The approximate EV/ACV multiple expanded all the same, from around 3.8 times on the June house anchor and Q1 balance sheet to about 4.9 times today. The stock has moved faster than the trailing software contract base.

Relative to the old June valuation bands of $9–13 bear, $15–22 base and $28–42 bull, my independently rebuilt SOTP moves the conservative and base centers higher. The reason is not the 20% stock-price rise. It is the combination of stronger ACV guidance, confirmed Ajax cash, additional enterprise validation, cost discipline and a more capital-efficient pipeline strategy. The optimistic ceiling does not move nearly as much because the biggest upside assumptions remain clinical and partnership assumptions that have not yet been de-risked.

Over the next year, ACV and cost control dominate. The company needs to show that Q2's strong contracting was not an isolated renewal quarter, commercial NDR needs to stop deteriorating, and Bunsen needs to move from early access to an actual commercial contribution.

Over three years, the critical variable is consolidated self-funding. The 2028 adjusted-EBITDA target matters more than any single milestone. A business with 10–15% ACV growth and high-70s software gross margins should eventually carry far more of its own R&D. Missing that target after the hosted conversion is substantially complete would say the cost structure, not accounting timing, is the problem.

Over five years, the decisive question is whether the platform keeps producing economically differentiated molecules. Software alone can justify a meaningful company. It probably cannot justify the most optimistic long-term equity outcomes unless the retained asset economics continue to pay. By then, investors should be able to judge the repeatability of the monetization engine from a much larger sample.

The company becomes materially more attractive when the software franchise can be purchased at a conservative software value while the pipeline is obtained for little or nothing. At today's price, that margin is not wide enough.

Bull reasons:

• Full-year ACV guidance of $218–228 million implies 10–15% growth despite reported software-revenue pressure from the hosted conversion, and Q2 ACV rose 27%.
• All top 20 pharmaceutical companies by 2024 revenue were software customers in 2025; top-20-pharma ACV rose 15.3% and gross commercial retention stayed at 96%.
• BMS is deploying the platform and Bunsen across its research organization, supplying enterprise-scale validation even though contract economics are undisclosed.
• Nimbus, Morphic, Ajax and Tectora show multiple independent cases in which Schrödinger-originated discovery work produced equity, milestone or downstream economics.
• At a 4.5–5-times ACV software valuation plus normalized cash, today's market price assigns little explicit value to the proprietary pipeline.

Bear reasons:

• Q2's $6 million net profit disappears after removing a $10 million Ajax milestone, and the quarter also contained $45.9 million of equity fair-value gains.
• Commercial NDR fell from 113% to 100% in 2025, showing that large-account strength has not translated into broad-based expansion.
• H1 2026 still consumed $48.9 million of operating cash and the company expects losses for several years, so pipeline optionality continues to have a real carrying cost.
• Stock-based compensation was $43 million in 2025, creating meaningful per-share dilution while consolidated owner earnings remain negative.
• The current price is slightly above the conservative SOTP and the base value falls to roughly the current share price if the assumed software multiple is cut from 5.0 times to 3.5 times.

Pre-mortem script one: by 2028, a combination of internally built pharma AI workflows, cheaper machine-learned simulation and broader offerings from scientific-software competitors reduces the amount customers are willing to pay specifically for premium Schrödinger compute. ACV growth falls below 5%, NDR drops below 95%, hosted gross margin settles near 70% rather than the targeted high 70s, and the market re-rates the software business from about five times ACV to three times. If ACV is then around $240 million, software EV would be only about $720 million. After another two years of cash burn, the stock could trade in the low teens even with no clinical disaster.

Pre-mortem script two: SGR-1505 and SGR-3515 fail to attract attractive partners during 2027, no co-founded asset produces a material realization, and collaboration milestones arrive later than expected. Management keeps enough R&D to preserve the platform, consuming roughly $80–100 million of annual cash, and misses the 2028 adjusted-EBITDA goal. The market removes most pipeline option value and values software at 3–3.5 times ACV. With a higher diluted share count, a price around $9–12 becomes plausible, representing roughly 40–50% permanent loss from today's level.

The first pre-mortem does not require FEP+ to “fail.” It only requires its economic scarcity to fall. The second does not require every drug to fail. It requires the portfolio to consume cash faster than realizations arrive.

The final judgment follows from those loss paths. Schrödinger owns a better business than its GAAP income statement suggests: the software franchise is growing on an ACV basis, the customer set is unusually strong, and the company has established a real record of turning computational discovery into downstream economics. The transition toward hosted contracts and externally financed drug vehicles is strategically sensible.

The price is less compelling than the business. At $19.02, a conservative SOTP produces roughly $18.7, while the base case reaches about $24.3. That creates attractive upside if execution continues, but no conservative margin of safety. The June-to-September rally has already capitalized part of the improvement through a higher EV/ACV multiple.

For an existing shareholder with a three-to-five-year horizon, holding through the accounting transition is defensible. For a new position, I would require either a materially lower price or another layer of fundamental de-risking: ACV above guidance without weakening retention, a priced Bunsen contribution, a credible partnership for a lead clinical program, or clear evidence that consolidated cash burn is converging toward the 2028 target.

【Company-profile scores】

• Fundamental quality: medium
• Growth: medium
• Moat: strong
• Financial soundness: strong
• Management credibility: medium
• Valuation attractiveness: medium
• Risk level: high
• Suitable investor type: long-term growth / event-driven

【Investment rating】

• Rating: Hold
• One-line thesis: ACV growth and asset monetization remain real, but normalized earnings are negative and the current price offers no conservative margin of safety.
• Ideal buy price: see dedicated line below.
• Acceptable hold price: 21–27 USD based on roughly ±15% around the $24.3 base SOTP, rounded conservatively.
• Clearly overvalued price: 37–42 USD, beginning about 10% above the $33.7 optimistic SOTP.
• Current-price classification: outside the three bands; the stock is above the ideal-buy zone but just below the acceptable-hold band.
• Whether to wait for a better price: yes for a new position. A price at or below roughly $15, provided FY2026 ACV remains at least $218 million and retention does not deteriorate materially, would create the required conservative margin. The opportunity cost is missing a re-rating driven by BMS/Bunsen adoption or a major pipeline partnership.
• Target holding horizon: 3–5 years
• Expected annualized return: conservative about -1%; base about 8%; optimistic about 21%, using a three-year convergence to scenario value and excluding tax, trading costs and further dilution.
• Max-loss risk: roughly 40–50% under the pre-mortem in which ACV slows to low single digits, the 2028 EBITDA target is missed, cash falls materially and software compresses toward 3–3.5 times ACV.
• Reassessment-trigger signals: FY2026 ACV below $218 million; commercial NDR below 95%; software gross margin below 70% for two consecutive quarters; liquid resources below $300 million without a visible self-funding path; or failure to achieve meaningful partnership progress for the lead proprietary clinical assets through 2027.

【Ideal Buy Price】13–15 USD Basis: at least approximately 20% below the conservative SOTP of about $18.7 per share, with the upper endpoint rounded to the nearest practical dollar and conditional on the ACV thesis remaining intact.

【Valuation Range】

• current: 19.02 (close as of 2026-09-11)
• bear (conservative · ideal buy zone): [13, 15]
• base (fair · acceptable hold zone): [21, 27]
• bull (optimistic · above the clearly-overvalued line): [37, 42]

The current price sits in the gap between the buy zone and the hold-value band. That gap is intentional: paying $19 rather than $15 sacrifices most of the conservative margin of safety, while the base SOTP still indicates upside. The rating is Hold rather than Cautious Buy because the downside case is driven by a valuation multiple that can compress without any catastrophic scientific failure.

Research uncertainties remain material. The largest is the BMS contract: scope is described, but duration, ACV and pricing are not. The second is Bunsen: it is in early access, and no standalone pricing, revenue or ACV is disclosed. The third is private-portfolio valuation: Tectora and several remaining private interests cannot be marked reliably from public information. The fourth is rNPV: early clinical programs have highly nonlinear probabilities, and small changes in assumed partnership economics produce large percentage changes in asset value. The fifth is the hosted transition: ACV gives a clean demand signal, but public disclosure is insufficient to reconstruct an audited counterfactual GAAP software-revenue series on a constant deployment basis.

Primary research relied principally on Schrödinger's Q2 2026 earnings release and SEC filing for current ACV, segment revenue, cash, Ajax accounting and guidance. The 2025 full-year release and annual filing supplied customer KPIs, hosted-mix data, the 2028 objectives, cash-flow history and portfolio disclosures. The BMS release supplied the scope of the new strategic software deployment, while the Tectora announcement supplied the newest spin-out structure and the company's latest “over $750 million” proceeds statement. Historical financial and capital-market work used company releases, SEC disclosures, the IPO announcement and verified historical price data. Peer work used current company disclosures from Certara, Recursion, Simulations Plus and XtalPi rather than inherited house ratings.

Other tickers mentioned

• CERT.US — closest listed benchmark for scientific software and model-informed drug-development economics.
• SLP.US — specialist simulation-software and services peer whose pending $18.50-per-share acquisition provides a transaction valuation anchor.
• RXRX.US — closest large listed platform-biotech comparison for valuing computational drug-discovery pipeline optionality rather than software.
• ABCL.US — AI-enabled biological-discovery platform mentioned as a broader sentiment peer, but not a close software comparable.
• ABSI.US — AI-enabled protein-design platform mentioned as a broader computational-biotech peer.
• RLAY.US — structure-based small-molecule drug developer relevant to the pipeline side of the competitive landscape.
• 02228.HK — XtalPi, a physics-, AI- and automation-driven discovery platform that illustrates the same milestone-versus-underlying-growth accounting problem.
• LLY.US — acquired Morphic and Ajax, crystallizing two important Schrödinger-linked equity realizations.
• BMY.US — longstanding drug-discovery collaborator and the customer behind the 2026 large-scale strategic software deployment.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.