Certara Deep Dive: A Biosimulation Software and Services Platform Embedded in Regulatory Workflows

Platform Formation

The company emerged because traditional drug development has two very expensive and slow links: the translation from preclinical to clinical work often relies on experience and trial and error, and many regulatory questions that could be answered earlier in models are left to later clinical trials. In its latest annual report, Certara defines MIDD as using biological and statistical models to support drug-development and commercialization decisions, with biosimulation as a core component that can support first-in-human dose prediction, clinical-trial simulation, drug-drug interaction prediction, label expansion, and other work. Simcyp was especially important because it gave Certara a truly regulatory-grade PBPK platform. In the 2025 annual report, the company says Simcyp has been developed for more than 20 years and is designed to generate "regulatory-quality" results that can be used directly in drug approval applications.

Compared with today, the core of the business model did not change at this stage, but its boundaries did. The constant was that software plus expert services remained the two engines. The change was that early software was more often specialist tooling sold to people who understood models. Today's Certara has extended those tools into clinical-data standardization, regulatory submission workflows, drug-discovery informatics, and AI-assisted writing. Early competition was also more fragmented, including single-point modeling software, in-house pharma teams, and academic tools. Over time, Certara gradually turned the question from "whose software is better" into "who can more reliably bring modeling results into regulatory and decision-making workflows." That is why it could slowly escape a single-tool price war.

PE Consolidation and Business Formation

After Arsenal took over in 2013, Certara entered a typical private-equity consolidation phase: acquisitions, product-line filling, services expansion, and platformization. In 2017, EQT acquired control of Certara for about 850 million USD, while Arsenal retained a minority stake. EQT's investment logic was straightforward: take a simulation and regulatory-technology platform that already had industry standing, continue consolidating it, and turn it into a public life-sciences software asset. The company later disclosed in its annual report that it had acquired 21 companies since 2013, including 14 related to software or technology. This number is critical. It explains why Certara's platform is so broad today, and it also explains why the balance sheet carries heavy goodwill and intangible assets, and why GAAP profit often looks unattractive.

The most important long-term effect of this phase was that Certara found its "hybrid" positioning. It retained a scientific and regulatory team deep enough to make software sell further. It did not go all the way into full-service CRO, and it did not behave like a pure software company that sells only seats and subscriptions. The benefit of this strategy is that it is easier to enter complex customer projects. The downside is that growth and margins will always be partly held back by services. For several years afterward, the market was willing to ignore this issue because the software narrative was more popular. In 2025–2026, the problem has been magnified again.

Listing, M&A Expansion, and Valuation Peak

Certara listed on Nasdaq in December 2020. The IPO priced at 23 USD, sold 29.055 million shares, and implied a market cap of about 4.56 billion USD at the offering price. The story it told the market at listing was clear: biosimulation would make drug development faster, cheaper, and more predictable, and Certara was one of the few platforms in the field with software, services, a customer base, and regulatory credibility. The market largely accepted that story at first. In March 2021, the company completed another block offering at 25 USD per share, followed by another selldown transaction at 31 USD per share in November, showing that the capital market was willing to keep measuring it on high-growth software coordinates.

Post-listing expansion also continued. In 2021, the company acquired Pinnacle 21 for about 310 million USD in cash and stock, bringing clinical-data standardization and pre-submission data compliance into the platform. In 2023, it acquired Vyasa, adding deep learning and unstructured biomedical information processing. In 2024, it completed the acquisition of Chemaxon, extending upstream into drug discovery and filling the cheminformatics link in the "design-make-test-analyze" workflow. Management stated clearly in the 2025 annual report that these transactions were intended to extend the platform from late-stage development into earlier discovery and data connectivity.

In hindsight, this stage was both the high point of Certara's business-boundary expansion and the starting point of distortion in the valuation story. The platform did become larger, and the software matrix did become more complete. But the capital market equated "a more complete platform" with "faster growth," and that inference has not fully materialized. Whether acquisitions such as Pinnacle 21, Chemaxon, and Vyasa can truly raise main-platform penetration and customer spend, rather than only assembling more revenue, remains under validation today.

New CEO, Active Slimming, and Revaluation

In January 2026, Jon Resnick formally became CEO and joined the board. He had spent a long period at IQVIA, most recently as president for the U.S. and Canada, and had also led IQVIA Real World Solutions. That resume naturally suggests two things. First, he is experienced in commercial execution for large life-sciences services and data organizations. Second, the board wanted someone who could turn the platform back into a sellable, deliverable, and scalable organization, rather than a scientific founder.

Resnick's first major move was to sell the regulatory-writing and medical-writing business, not to talk up growth. In April 2026, the company announced the sale of this business to Veristat for consideration of up to 135 million USD. In 2025, the business generated about 50 million USD of revenue and about 17 million USD of adjusted EBITDA and involved about 220 employees. During the May Q1 report, management said the transaction had been completed and updated 2026 guidance to 395 million to 405 million USD, including residual first-half revenue from the divested business. This move matters because it shows that the new CEO's diagnosis is to cut away the parts closest to billable hours and less connected with the core MIDD platform, then move resources toward core software and high-barrier services, rather than to add services for their own sake.

Whether this step changes the company's fate cannot yet be concluded. It at least changes the narrative. Certara is no longer pretending it can do everything. It is acknowledging that its services boundary had stretched too widely and hurt growth quality and organizational efficiency. The problem is that the market will not immediately grant a higher valuation just because one business was cut. What the market really needs to see is whether software can grow faster, services can become more stable, and the organization can operate with less friction after the divestiture. In 2026, Certara has entered a stage in which it must speak through quarterly data.

Longitudinal Financial Review

If one only looks at full-year 2025, Certara still looks like a company of decent quality. If the lens moves closer to Q4 2025 and Q1 2026, it looks like a company losing speed. The real issue is that both appearances are true at the same time.

The table reflects two facts that coexist. First, total revenue grew steadily from 2023 to 2025, and operating cash flow stayed positive. Second, GAAP profit has long been distorted by acquisition amortization, stock-based compensation, and contingent-consideration remeasurement, so the apparent improvement in profitability has lagged cash flow. Based on 2025 revenue of 418.8 million USD and costs of 161.1 million USD, blended gross margin was about 61.5%, slightly higher than in 2024. The 2025 adjusted EBITDA margin was about 32.1%, also higher than in 2024. The problem is that growth increasingly depends on the software leg alone, while services are still larger. In 2025, software revenue was 183.3 million USD, up about 17.7% year over year; services revenue was 235.6 million USD, up only about 2.7%. In Q4 2025, software still grew 10%, while services turned to -1%. In Q1 2026, software was +7% and services -4%. The real picture of the slowdown is therefore that the better part is still growing, while the larger part has started to drag, not that every product across the company is deteriorating together.

Cash-flow quality is much better than GAAP profit. In 2025, operating cash flow was 96.3 million USD, compared with a full-year GAAP net loss of only -1.6 million USD. Relative to adjusted net income of 70.9 million USD, cash conversion was about 1.36x. For an acquisition-driven software-and-services hybrid like Certara, low GAAP net income is normal, mainly because of amortization of acquired intangible assets, stock-based compensation, and the 7.4 million USD fair-value change in Vyasa contingent consideration in Q1 2026. What matters is whether cash flow grows with revenue and whether it is increasingly consumed by capitalized investment, not whether the company earns GAAP profit. Based on current disclosure, operating cash flow remains healthy.

The balance sheet has two sides. On one hand, at the end of Q1 2026 the company held 149.5 million USD of cash and had a term-loan balance of 294.8 million USD, so leverage was not dangerously high. On the other hand, this is a platform company that has made acquisitions for more than a decade, with heavy goodwill and intangible assets on the balance sheet. If some acquired categories fail to deliver cross-selling or growth, that will show up through impairments or weaker-than-expected margins. In addition, the company started using a 100 million USD repurchase authorization in 2025 and had repurchased 82.6 million USD by the end of Q1 2026, including 40.0 million USD in that quarter alone. This indicates that management thinks the stock is cheap, and also that when growth is not fast enough, the company has begun to rely more on capital allocation to support shareholder returns.

From a return-on-capital perspective, Certara looks more like a software asset that can generate cash but has unstable growth and accounting profit, rather than a traditional services provider that requires heavy asset investment. The company does not disclose a clear split between maintenance and growth capex, which complicates valuation. But given the business form, the "maintenance spending" that should really be deducted from profit sits more in ongoing R&D, product upgrades, and some capitalized software investment than in data centers and equipment. In other words, Certara's true owner earnings are higher than GAAP net income but lower than unadjusted operating cash flow. For this kind of company, watching cash flow is more useful than watching EPS.

Share Price and Valuation History

Certara's valuation history is almost a condensed history of the ebbing life-sciences software narrative. At the end of 2020, the market was willing to value it at nearly 4.6 billion USD at IPO because imagination around "digital drug R&D," "software replacing trial and error," and "high-gross-margin life-sciences SaaS" was very generous. The two block or secondary offerings in 2021, at 25 USD and 31 USD, further show that investors then viewed it as a growth stock capable of continuing to expand its valuation center.

The subsequent valuation correction came in three steps. The first was the financing headwind. Slower biotech financing naturally hits small and mid-sized biotech companies and project-based services demand first. The second was exposure of the company's revenue structure: software growth was good, but services were a larger share, so overall growth could not lift. The third was that the market began to evaluate it on stricter coordinates. If full-year growth guidance is only 0%–4%, it looks more like a low-growth, cash-generative software hybrid than high-growth SaaS. Arsenal's 2022 block purchase from EQT at 15 USD per share itself marked a lower valuation center. By June 2026, the 5.27 USD share price had cut that center down by more than half again.

Using the latest 2026 framework, Certara's current market cap is about 831 million USD. Based on cash and debt at the end of Q1 2026, enterprise value is about 977 million USD. Against the midpoint of 2026 revenue guidance of about 400 million USD, current P/S is about 2.1x and EV/Sales about 2.4x. Based on a midpoint adjusted EBITDA margin of 31%, EV/adjusted EBITDA is about 7.9x. This pricing is now closer to an asset with slower growth but still decent cash flow, and no longer resembles the 2020–2021 "software story first" valuation. Because I did not rebuild a full daily historical valuation percentile series in this one-off study, I will not write a precise percentile. But the direction is clear: the current valuation is materially below the IPO, below all 2021 refinancing prices, and below Arsenal's 2022 entry price.

Business Model and Moat

Revenue Structure

Certara's revenue structure looks simple but is layered in practice. The software side can be broadly divided into four layers: first, core biosimulation tools represented by Simcyp and Phoenix; second, drug-discovery informatics represented by D360 and Chemaxon; third, data standardization and automation represented by Pinnacle 21 and Metadata Repository; fourth, regulatory submission and writing tools represented by GlobalSubmit and CoAuthor. The services side is built around QSP, pharmacokinetic/pharmacodynamic modeling, clinical pharmacology, data science, regulatory strategy, and regulatory operations. The company also states clearly that its software can be delivered through on-premise licensing or cloud-based subscription.

The true profit source remains software, not services. Full-year data already provide the evidence: software revenue growth in 2025 was much higher than services growth. By Q1 2026, software bookings grew 20% and software net retention was 106%, while services bookings fell 14%. This shows that expansion and additional purchasing are still happening within the installed software base, while services are more affected by budget timing, customer execution, and project timing. The sale of the regulatory-writing and medical-writing business is essentially an acknowledgment that not every service belongs on this platform long term. The writing business could contribute EBITDA, but it was closer to a labor-intensive organization and less like core MIDD services that can drive software penetration.

The customer base is relatively diversified. At the end of 2025, no single customer contributed more than 10% of revenue, and the top ten customers together contributed 24%, slightly below 27% in the prior year. This means the company does not face an obvious problem where the loss of one large customer would cause severe damage. What it really worries about is budgets inside the same customer shifting from "use software plus services to validate questions earlier" back to "cut budgets first, do fewer models, and only run necessary projects," rather than customer concentration.

Cost Structure and Operating Leverage

This company naturally has operating leverage, but it does not appear automatically. More software means better margins; more services mean steadier revenue but lower margins. In 2025, the company's blended gross margin was slightly above 60% and already better than in 2024. But pressure in Q4 and Q1 also proved that once the revenue mix shifts toward services or services utilization falls, margins are affected immediately. Costs did not expand materially in Q1 2026. The real drag on profit was services weakness, business-mix transition, and extra expense from the Vyasa contingent-consideration remeasurement.

The hardest cost to compress is people, not servers. Modeling scientists, clinical-pharmacology experts, regulatory operations teams, and data teams are all high-paid knowledge workers. For Certara, software scale can dilute R&D and sales costs, but if services demand suddenly slows, the organization cannot cut costs as quickly as a pure software company. This is also why the new CEO's first cut was selling the writing business, rather than continuing to stack services revenue.

Moat

Certara has four real moats.

The first is regulatory embedding. It rests on the fact that its software is used by 20 global regulatory agencies and, according to the company's internal statistics, more than 90% of FDA new drug approvals from 2014 to 2025 involved its customers, rather than on the company simply saying regulators like it. Add the FDA's MIDD paired-meeting mechanism, PBPK submission-format guidance, and the EMA's formal documents on PBPK platform qualification and reporting requirements, and the essence of this moat is that models and results must be acceptable to regulators. It is not a license. Regulatory-grade explainability and reproducibility do not automatically lose value because large models become popular.

The second is deep switching cost. Simcyp, Phoenix, and Pinnacle 21 are tools embedded in R&D processes, data workflows, and submission workflows, not one-time tools. Customers are buying not only a software package, but modeling habits, historical model libraries, internal SOPs, training systems, and review-communication experience. Replacing them requires rebuilding models as well as internal validation and regulatory-interaction practices. This switching cost is harder and slower than replacing an ordinary enterprise software system.

The third is mutual reinforcement between software and expert services. Pure services companies find it hard to turn services into products, while pure software companies often lack the last-mile regulatory and project-application capability. Certara's advantage is that it can sell software and also send scientists to apply the software to customer projects, then feed project experience back into products. The annual report directly says the company uses validated software applications to deliver technology-enabled services. This is where it is harder to replace than many single-point software vendors.

The fourth is platform coverage. Chemaxon and D360 extend into discovery. Simcyp and Phoenix sit in development. Pinnacle 21 and Metadata Repository manage data governance. GlobalSubmit and CoAuthor manage submission. Customers do not necessarily buy the full suite at once, but once the platform is deep, add-on sales are easier than starting from zero. The 106% software net retention in Q1 2026 is the most direct evidence.

Two moats that the market likes to discuss need cooling down. First, the network effect is not strong. More customers do not significantly increase the value of the product to each individual customer the way an exchange or social platform would. Second, AI itself is not a moat. The company can embed AI into the toolchain, but the annual report is also clear that AI lowers entry barriers and attracts larger technology players. AI is therefore more like an amplifier or eroder of the moat, not the moat itself.

Management and Governance

Jon Resnick's arrival marks a shift from the "scientist-led platform consolidation phase" to an "operator CEO-led organizational rebuilding phase." He comes from IQVIA and has sales, delivery, and organizational experience in large-scale life-sciences services and data businesses. The board also made parallel changes in 2026: Arjun Bedi joined the board, while Rosemary Crane and Cynthia Collins stepped down. This suggests that the board is actively refreshing itself rather than waiting passively.

The largest governance variable is Arsenal, not insider ownership. According to the 2026 proxy filing, Arsenal held 36.346 million shares as of 2026-03-20, or about 23.71%, and remained a decisive significant shareholder. Its director-nomination rights can remain at two seats if ownership thresholds are met and fall to one seat in the 6%–12% ownership range. More importantly, in April 2025 Arsenal extended the original transfer-restriction agreement, which had been scheduled to run through the end of 2024, to 2026-04-14. This means the lockup had expired by the report date, and the market had not seen an updated Arsenal 13D amendment. The implication is direct: whether future selldowns happen, and at what pace, will be part of the governance discount.

Industry and Cycle Analysis

Certara sits at the intersection of three industries, not inside a single industry: life-sciences R&D software, model-informed drug-development services, and regulated development-process digitization. The benefit of this intersection is that demand does not come from a single budget bucket. The downside is that it is affected by software valuation logic, biotech R&D budget cycles, and changes in regulatory thinking.

The industry backdrop is clearly favorable. The FDA launched the MIDD pilot meeting program in 2018 and updated it into the MIDD Paired Meeting Program in 2026. The FDA issued PBPK analysis submission-format guidance in 2019 and guidance on PBPK applications in biopharmaceutics for oral drug products in 2020. The EMA also has formal guidance on PBPK modeling and simulation reporting, explicitly applying to both commercial platforms and internally developed platforms. Once regulatory language moves from "can be discussed" to "has a formal submission template," industry demand moves from "advanced method" to "process infrastructure."

But however good the industry is, it is not acyclical. Certara itself states in the annual report that global pharmaceutical R&D spending was about 290 billion USD in 2025 and that more than 6,800 biopharma companies had active R&D pipelines. This of course indicates a large long-term demand pool. The problem is that Certara's marginal customers are more likely to be small and mid-sized biotech and drug-development companies that may delay projects when financing conditions change, rather than the cash-rich large pharma companies. On the Q1 2026 call, management said all software customer tiers performed at or above plan, while services outside tier-one customers remained soft, and that the company expected tier-two and tier-three markets to recover during the year as the financing environment improved. Put more plainly, this is a "long-term non-cyclical, short-term cyclical" market. Regulatory acceptance keeps rising over the long run, but quarterly orders are still affected by biotech financing, project-initiation cadence, and internal customer budget control.

From the perspective of profit pools, the most profitable position is held by whoever controls R&D decision tools that customers must use and keep upgrading, not whoever has the most people. This is why software-side value is higher than services-side value. Traditional CROs earn revenue through headcount and execution. They can be large, but margins are not necessarily high. Modeling and regulatory-grade software earn money through process embedding and standardized output. They need not be the largest, but they often command better valuations. Certara's difficulty is that it has not yet shifted its revenue structure enough for the capital market to value it fully as software-pool exposure. It is not in the wrong profit pool; its mix is not yet clean enough.

Policy and regulation also bring a reflexive risk that is often overlooked. Certara's annual report lists U.S. federal agency staffing reductions and HHS/FDA reorganization as risk factors, noting that they may affect FDA and HHS staffing, procedures, decision cadence, government contracts, and the broader market environment. For a company deeply embedded in regulatory workflows, regulatory support is a demand driver, while contraction inside the regulatory system itself can also create near-term friction. It may not change the long-term logic, but it can affect execution over one to two fiscal years.

Horizontal Competitor Analysis

Certara has direct comparables, but no single company maps it completely. A more accurate approach is to split it into two halves: software and modeling platforms, compared with Simulations Plus and Schrödinger; and services and penetration into pharma workflows, compared with IQVIA and ICON. The benefit is that this clarifies Certara's real position: it is neither the purest software company nor the heaviest services company.

Market caps in the table use values around the 2026-06-11 close. Revenue and growth use each company's latest public guidance or most recent full-year data. Valuations are research estimates based on the above data.

Simulations Plus is the closest listed company to the scientific-software character of Certara's core, but it is far smaller and cleaner. Its FY2026 guidance is also only 0%–4% growth, which is interesting: it shows that Certara is not the only company facing slower growth in life-sciences modeling software. The difference is that SLP's Q2 FY2026 revenue still grew 8%, both software and services grew, software was about 60% of revenue, and adjusted EBITDA margin reached 36%. It looks more like a small, pure scientific-software company. The market is willing to give it a higher P/S because the structure is cleaner and the services burden lighter. Certara's advantage is broader platform coverage and deeper regulatory embedding. SLP's advantage is that it looks more like a true software company.

Schrödinger represents another path: pushing the computational platform deeper into the front end of drug discovery and AI co-development, trading technology premium for greater long-term imagination. In Q1 2026, its ACV grew 12%, but software revenue fell 21% year over year because the company was actively shifting its licensing model to hosted licensing. At the same time, drug-discovery collaboration revenue rose sharply and cash reserves remained at 406 million USD. Its valuation rests on technology route, collaboration milestones, and the AI narrative, not current profit. Compared with Schrödinger, Certara's biggest difference is stronger regulatory implementation and more visible cash flow. Its biggest disadvantage is that the story is less exciting and the services drag is more obvious.

IQVIA is Certara's new CEO's former employer and the best services reference point for understanding Certara. IQVIA's advantages are scale, customer coverage, data assets, and global execution systems. In Q1 2026, revenue was 4.151 billion USD, adjusted EBITDA was 932 million USD, cash flow was strong, and repurchases continued. It shows one thing: life-sciences services and technology platforms can be very profitable, but only if they have extremely strong distribution and scale advantages. Certara cannot copy IQVIA at scale, so it must win through deep specialization in a high-barrier smaller market, rather than trying to become a smaller IQVIA. Resnick's arrival may improve commercial execution, but it does not mean the company will naturally grow into another IQVIA.

ICON provides another warning. It is a purer full-service CRO, but in 2026 it had to delay its annual report and disclose that 2023 and 2024 revenue may have been overstated by less than 2% because of revenue-recognition investigation and financial-control issues, visibly hurting both the stock and the narrative. ICON and Certara are not directly in the same market. It is included because it reminds investors that when governance, revenue recognition, and project-execution risks in services businesses go wrong, valuation drawdowns can be very fast. By comparison, Certara's governance issue is the exit pace of its large shareholder and whether the new CEO's pivot can be delivered, not accounting quality. The software moat can make it more resilient than traditional CROs, but the high services share still makes the market more sensitive.

Taken together, Certara's niche is clear. It is one of the leaders in biosimulation and regulatory-grade modeling toolchains, positioned between purer software vendors and larger CROs. The gap it fills is the segment that connects modeling, data standardization, and regulatory-grade submission. What it most directly takes share from is inefficient internal pharma trial and error and some unnecessary clinical-trial cost. The profit pools most likely to attack it are platform players that build stronger front-end simulation while turning AI into a low-cost assistant, as well as partial in-house substitution as customers strengthen internal capabilities. If technical substitution occurs, pressure will first fall on more standardized services and writing work that can be assisted by AI, not on regulatory-grade platforms such as Simcyp. This also explains why Certara sold the writing business first.

Current Fundamentals and Bull-Bear Debate

What Has Truly Changed in Recent Quarters

The most important change in recent quarters can be summarized in one sentence: software has not broken, services loosened first. In Q4 2025, total company revenue was 103.6 million USD, up 3% year over year; software was 46.4 million USD, up 10%, while services were 57.3 million USD, down 1%. Software bookings also fell 6% in Q4, which management attributed to external factors and execution challenges, while services bookings rose 17%. The year-end 2025 problem therefore looked more like software booking timing had wobbled while services were still taking orders. By Q1 2026, the picture had reversed: software revenue was 49.7 million USD, up 7%; software bookings grew 20%; software net retention was 106%; services revenue was 57.2 million USD, down 4%; and services bookings fell 14%. In other words, software regained footing, while services began to show more real weakness.

Profit also reflects a transition period. Q1 2026 adjusted EBITDA was 31.7 million USD, down year over year, with a 30% margin. GAAP net loss was 8.8 million USD, mainly because the fair-value change in contingent consideration related to the Vyasa acquisition added 7.4 million USD of expense. This quarter should not be read simply as "the business collapsed." A more accurate reading is that the core software business remained reasonably stable, services and organizational change pressured profit, and GAAP numbers were made more volatile by legacy M&A items.

What the Market Is Trading Now

The market is trading a growth penalty, not an AI premium. After the company gave full-year guidance of 0%–4% in February 2026, the valuation logic shifted from "growth" to "repair." After the May sale of the regulatory-writing business and updated full-year guidance, management emphasized that core business growth expectations had not worsened, but the market was not more reassured because services weakness had been further confirmed. After the Q1 report, multiple sell-side firms lowered target prices. The market's core reaction was that even after low-quality revenue was sold, core growth had not immediately improved meaningfully. It was not simply that selling a business made revenue smaller.

Bull-Bear Debate

The bull case has three real core arguments. First, software demand has not collapsed: Q1 2026 software revenue was +7%, bookings +20%, net retention 106%, and management said all software customer tiers performed at or above plan. Second, the services problem may be cyclical and execution-related: management explicitly linked improvement in tier-two and tier-three markets to the stronger financing environment during the first four months of 2026, suggesting it does not believe the demand pool has permanently shrunk. Third, the active sale of the regulatory-writing and medical-writing business will make the revenue structure closer to a high-quality core platform, increasing the chance that the capital market eventually prices it as "software plus high-barrier services" rather than a mixed services company.

The bear evidence is also concrete. First, full-year core growth remains only 0%–4%, and management explicitly says services growth will be near the low end of the range, with first-half performance possibly even below the lower end. Second, the company has made many acquisitions in recent years and stretched its boundaries wider, yet total growth is still only low-single-digit today, suggesting that platform expansion has not automatically turned into higher growth. Third, AI is not a one-way positive. The company itself acknowledges in the annual report that AI may allow new entrants and larger technology companies to enter relevant markets at lower cost. Fourth, Arsenal's transfer restrictions expired on 2026-04-14, while its stake remains close to 24%; any future selldown could pressure the stock.

Valuation Analysis

Historical Valuation

The most defensible conclusion today is that the market no longer treats Certara as a high-growth software stock, not that it sits at a specific historical percentile. On price, it is clearly below the 23 USD IPO price, below the 2021 offering and selldown prices of 25 USD and 31 USD, and below Arsenal's 2022 transaction price of 15 USD. On valuation coordinates, based on the midpoint of 2026 guidance, current P/S is about 2.1x, EV/Sales about 2.4x, and EV/adjusted EBITDA about 7.9x. This is now closer to a low-growth, high-cash-conversion software-services hybrid than to a life-sciences software stock in the SaaS bull market.

Peer Valuation

Compared with peers, Certara is much cheaper than Simulations Plus and also cheaper than Schrödinger. Compared with larger and more services-oriented companies such as IQVIA and ICON, its price-to-sales ratio is not absurdly low. This discount is basically reasonable because the market is applying half punishment and half collateral damage. The punishment comes from low growth and the services drag. The collateral damage comes from the market extrapolating "services weakness" into "the software moat is failing too." If the next two or three quarters prove that software net retention can stabilize around 105% and services bookings turn positive, the discount can narrow. If services continue to drag and platform restructuring does not translate into higher software penetration, the discount will widen instead.

Cash-Flow Look-Through

Look at cash flow first, then profit. In 2025, operating cash flow was 96.3 million USD against near-breakeven GAAP net income. This OCF/net-income ratio does not explain much for Certara because the GAAP denominator is structurally compressed by acquisition amortization and non-cash items. Relative to adjusted net income of 70.9 million USD, cash conversion was good. The problem is that the company does not disclose an exact split between maintenance capex and growth capex, so all "owner earnings" estimates can only be approximate. My conservative treatment is to apply a moderate discount to 2025 operating cash flow and treat normalized owner earnings as about 85 million to 90 million USD. Against the current 831 million USD market cap, owner-earnings yield is roughly 10%–11%. That yield is not low in isolation, but it rests on the premise that services do not continue to deteriorate, software remains stable, and future acquisitions do not produce obvious impairments.

Absolute Valuation

For Certara, I would not use a single DCF, and I would not rely only on PE. The more appropriate approach is to consider three items together: EV/Sales, EV/adjusted EBITDA, and owner-earnings yield. The reason is simple. On one hand, it has software attributes, so price-to-sales cannot be ignored. On the other, it already has real profit and cash flow, so it cannot be valued with a "TAM only, no profit" framework.

Based on these assumptions, my estimated intrinsic values are roughly 5.1 USD per share in the conservative case, 6.1 USD in the base case, and 8.6 USD in the bull case. This is a research framework that separates growth, margin, and multiples, not investment advice. The underlying data supporting these estimates are the current stock price and market cap, the latest 2026 guidance, the Q1 2026 cash and debt structure, and current peer valuations.

Expectation Gap Analysis

Current market expectations are not high. The market assumes Certara must go through at least several quarters of low growth and restructuring pain before it deserves a valuation repair discussion. What can create an expectation gap is two hard metrics, not another AI narrative. The first is whether software net retention can stay at 105%+ for consecutive periods and drive simultaneous improvement in software revenue and bookings. The second is whether services bookings can end year-over-year declines. If both appear at the same time, the market will again believe this is a software platform dragged by services, rather than a services hybrid whose software is also near a ceiling. Conversely, if software net retention falls back near 100%, the entire investment logic must be rewritten.

Margin of Safety Review

At the current price, the margin of safety relative to conservative implied value is not obvious. Based on my conservative intrinsic value of 5.1 USD, the current price of 5.27 USD offers almost no discount. The most fragile assumption is that services can improve toward the high end of the range in H2 2026. If that assumption is cut by 30%, revenue and margin in the base case both need to be lowered, and the base value would fall from about 6.1 USD to about 5.2–5.4 USD.

If earnings show zero growth over the next three years, I think the theoretical return at the current price is closer to owner-earnings yield than EPS growth. Based on an owner-earnings yield of about 10%–11%, the static return is higher than the U.S. 10-year Treasury yield of about 4.55% on 2026-06-10. But that comparison has one premise: the market must be willing to maintain the current valuation multiple rather than compress it further because of zero growth. For that reason, I do not view it as a stock with a sufficient margin of safety. It is the type of asset where quality has become much cheaper, but not cheap enough to earn money despite mistakes.

Conclusion on margin-of-safety sufficiency: not obvious.

Risk Analysis

I think Certara currently has five risks that could cause permanent capital loss.

The first is services weakness shifting from cyclical to structural. Probability is medium and impact high. Evidence has already appeared: Q4 2025 services revenue -1%, Q1 2026 services revenue -4%, services bookings -14%, and management explicitly said full-year services growth would be near the low end of the 0%–4% range. If services bookings remain negative over the next two quarters and software growth cannot expand, total company growth will be pinned at low-single digits for a long time. The market will then treat Certara as a slow-growth services provider rather than a software platform, and the valuation center will keep moving lower. The transmission path is insufficient orders → utilization pressure → EBITDA margin falling below 30% → multiple compression.

The second is that the market narrates AI's effect on the company backwards. Probability is medium and impact high. The company is putting AI into Certara.AI, CoAuthor, and modeling workflows, while also saying in the annual report that AI may lower entry barriers and attract new players or even large technology companies into relevant settings. The truly dangerous scenario for Certara is that high-standard software remains necessary, but pricing for some standardized services, writing, and data-organization work falls quickly, leaving software without price increases while services decline in price. A full AI replacement of modeling is not realistic in the short term. The transmission path is AI reducing unit prices for labor links → services gross margin pressure → lower platform cross-selling value → the market stops paying a premium for "software plus services."

The third is Arsenal's selldown and governance discount. Probability is medium and impact medium-high. According to the proxy filing, Arsenal owns about 23.71%, and the transfer restrictions expired on 2026-04-14. As of the research date, I had not seen an updated Arsenal ownership-change filing. Even without fundamental deterioration, the market starting to trade the exit path of a large shareholder would be enough to pressure a small- to mid-cap stock for a long time. The transmission path is rising expected overhang → slower valuation-repair pace → any negative news amplified.

The fourth is M&A aftereffects. Probability is medium and impact medium. Since 2013, the company has completed 21 acquisitions, including 14 related to software or technology. In Q1 2026, Vyasa contingent-consideration remeasurement created extra expense. Acquisitions can fill the platform, but they also mean that if any business underperforms expectations, the impact may appear through intangible-amortization pressure, integration friction, goodwill impairment, or consideration adjustments. The transmission path is poor integration → one-off items erode profitability → investors lose patience with adjusted metrics.

The fifth is friction inside the regulatory system itself. Probability is medium and impact medium. Certara's annual report explicitly lists HHS/FDA reorganization and staffing reductions as risks because they may affect review workflows, government contracts, and the broader market environment. For an ordinary CRO, this risk is more macro background. For Certara, it is part of the demand logic. The more regulation encourages MIDD, the more the company benefits. If the regulatory system itself falls into reorganization friction first, the pace of validating new methods and driving customer adoption will slow.

Catalysts and Tracking Metrics

Positive Catalysts

There are four practical positive catalysts. First, software net retention continues to stay above 105% while software bookings grow double digits for two consecutive quarters, showing software is truly resuming expansion rather than merely defensively stabilizing. Second, services bookings turn positive, especially with recovery among tier-two and tier-three customers, proving that the 2026 growth slowdown is more about the financing cycle and execution adjustment. Third, resource reallocation after the sale of the writing business begins to show up as higher software cross-selling, especially linkage among Chemaxon, D360, Pinnacle 21, and Simcyp/Phoenix. Fourth, repurchases continue and accelerate in clearly undervalued zones rather than being executed mechanically.

Negative Catalysts

The negative catalysts are equally clear. First, another cut to full-year guidance, even only on the services side, would make the market assume 0%–4% is the norm rather than a transition. Second, software net retention falling back to 100%–102% would directly shake moat-based pricing. Third, Arsenal files new selldown-related disclosures. Fourth, AI messaging keeps increasing while measurable product revenue does not appear, leading the market to view the story as narrative offset against reality. Fifth, any new noise related to M&A integration, consideration remeasurement, or financial quality would reduce confidence in adjusted metrics.

Tracking Dashboard

† Arsenal's stake uses the latest public figure disclosed in the 2026-03-20 proxy filing. The dashboard's "normal range" and "warning threshold" are a research tracking framework, not formal company guidance. Underlying data come from the company's Q1 report, annual report, proxy filing, and this report's valuation framework.

Among these metrics, only three are most worth watching: software net retention, services bookings, and EBITDA margin. The first answers whether the moat still exists. The second answers whether the issue is cyclical or structural. The third answers whether the hybrid model will increasingly resemble an inefficient services company.

Cross-Sectional and Longitudinal Synthesis

What Certara's history over the past decade-plus truly proves is that it did stitch a previously fragmented market for models, data, and regulatory tools into a platform capable of serving regulatory-grade decisions. It did not merely tell a story. From assembling Tripos, Pharsight, and Simcyp into an initial structure, to continuing acquisition-led capability filling in the PE era, to acquiring Pinnacle 21, Vyasa, and Chemaxon after the IPO, it has proved platform-integration ability, understanding of regulatory workflows, and the ability to sell expert services and software together. Its success came from a series of correct but costly choices, not luck and not industry tailwinds alone: continuous acquisitions, retaining services capability, and avoiding the shortcut of becoming "pure software."

Those success factors broadly remain today, but marginal changes have emerged. The long-term trend of regulatory acceptance of MIDD has not reversed, and the product matrix is more complete. From 2025 to Q1 2026, the software side remained the healthier part of the company. What has really changed is the capital market's patience. In the past, the market was willing to pay in advance for a platform that kept becoming more complete. Today, it demands evidence of restored growth before granting valuation. Certara's weakness, at least for now, is that its organizational and commercial structure prevents the value of high-quality software from fully flowing through the consolidated financials, not that the technology has been replaced. Services are too large, making it hard for good software alone to lift the whole company's growth curve.

Horizontal comparison further shows that Certara's real advantages are specific. It has more technology content and regulatory embedding than traditional CROs, sits closer to real projects and regulatory submissions than purer software vendors, and has more visible cash flow and real-world implementation than front-end AI drug-discovery platforms. Its weaknesses are equally specific. It is not as clean as Simulations Plus, not as rich in technology-premium imagination as Schrödinger, and not as dominant in global commercial execution radius as IQVIA. This position means it will not naturally receive the highest multiple, but it should not be treated as the worst asset either. It is a company that needs to prove it deserves to be seen as a platform again.

I think the market is most likely to misjudge two things. One is extrapolating services problems to the whole company and ignoring that software bookings and net retention have stabilized first. The other is seeing AI only as an incremental positive while missing that it may also accelerate pricing pressure in some services links. Therefore, the key variable over the next 1 year is whether services can stop falling. The key variable over the next 3 years is whether the software platform can restore mid- to high-single-digit growth. The key variable over the next 5 years is whether the company can transform itself from a hybrid asset into a software-centered platform asset. If these three variables materialize in sequence, Certara will be more investable than it is today. If the first layer cannot even materialize, today's seemingly inexpensive price will still not be safe enough.

Bull and Bear Arguments

Bull arguments:

• The software side reaccelerated in Q1 2026, with revenue +7%, bookings +20%, and net retention of 106%, showing that the core installed base still has expansion capacity.

• The company's software is used by 20 global regulatory agencies, and according to internal statistics more than 90% of FDA new drug approvals from 2014 to 2025 involved its customers, showing deep regulatory embedding.

• The active sale of the regulatory-writing and medical-writing business acknowledges and corrects the problem of overly broad services. The direction is right.

• 2025 operating cash flow was 96.3 million USD, and the estimated owner-earnings yield at the current market cap is around 10%, so valuation is no longer expensive.

• The new CEO comes from IQVIA, and the board is visibly pushing the company from integration phase into execution phase.

Bear arguments:

• Full-year core growth is still only 0%–4%, and services growth is explicitly near the low end. Low growth is the company's own guidance, not market imagination.

• Services revenue and services bookings have both shown repeated weakness. If this continues, the overall revenue structure will be hard for the market to price as a software platform again.

• The company assembled the platform through 21 acquisitions over the past decade-plus, creating high integration complexity and keeping GAAP earnings quality under pressure from M&A aftereffects.

• The company itself acknowledges that AI is both an opportunity and a new source of competition, especially with potential pricing pressure on standardized services work.

• Arsenal owns close to 24%, and transfer restrictions have expired, so future selldown cadence will continue to weigh on risk appetite.

Pre-Mortem: Where I Could Be Wrong

The first likely loss scenario is that the services problem is structural rather than cyclical. Suppose by mid-2027, despite an improved biotech financing environment, Certara's services bookings remain negative year over year for four consecutive quarters, while software net retention falls from 106% to 101%–102%. That would mean platform cross-selling is not working and customers are concentrating more budget on a few core software seats and internal teams. In that case, 2027 revenue growth could be only 1%–2%, adjusted EBITDA margin could fall to 27%–28%, and the market could compress EV/EBITDA from about 7.9x today to 6x. The stock could fall into the 2.5–3.0 USD range, implying a three-year loss of more than 40%–50%. Related clues have already appeared in Q1 2026: services weakness, negative services bookings, and a full-year services outlook near the low end of the range.

The second scenario is that AI improves tools but crushes pricing in the middle-layer services. Suppose in 2027–2028 platforms such as Schrödinger keep extending agentic AI and discovery workflows downstream, while purer software vendors such as Simulations Plus also use faster product iteration to take customers. At the same time, pharma companies begin moving parts of regulatory writing, data organization, and model pre-processing to lower-cost AI-assisted workflows. Certara's own software can still sell, but some high-margin services can no longer maintain prior pricing, forcing the company to protect project entry points with lower prices. Revenue may not collapse, but margins would be worse than the market currently assumes. Valuation would not rise simply because "the company has AI" and could instead step lower because AI erodes labor-based services. This scenario has real basis because the annual report explicitly acknowledges that AI may allow new entrants and larger technology companies to enter its customer settings.

Final Research Conclusion

The most interesting research question about Certara today is what price is appropriate to own a quality asset after growth has stalled. Whether it is a quality asset is relatively clear: it is. If viewed as a pure software company, the current price is already cheap. If viewed as a hybrid services company that can only grow low-single digits long term, the current price is not especially cheap. My judgment is closer to the middle. Certara's software and regulatory embedding mean it should not be valued like the weakest services company. But before services regain footing, it also does not deserve the life-sciences platform premium of 2020–2021.

This therefore looks more like a name entering a trackable range than a buy-with-eyes-closed opportunity. The current price is low enough for long-term investors to put it back near the front of the watchlist. But it still needs one services stabilization and one further price concession to offer a truly attractive margin of safety. My biggest concern is that the company has strong software assets but remains unable to change the overall structure enough to look more like a software platform, rather than that regulators suddenly stop recognizing modeling. The positive conditions that would change my mind are consecutive validation of software net retention and software bookings, a clear turn in services bookings, and resource allocation after the sale of the writing business starting to show up in overall growth. The negative conditions that would overturn my view are the software side also slowing at the same time, or Arsenal's exit overhang combining with deteriorating fundamentals to create a double hit.

【Company Profile Scores】

• Fundamental quality: Medium-high

• Growth: Medium

• Moat: Medium-strong

• Financial resilience: Medium

• Management credibility: Medium

• Valuation appeal: Medium

• Risk level: Medium-high

• Suitable investor type: Long-term growth / value stock

【Investment Rating】

• Rating: Watch

• One-line investment thesis: The software moat remains intact, but the services drag and restructuring execution still need quarterly proof.

• Three-tier price signals: Ideal buy price: see the next line

• Acceptable hold price: 5.2–7.0 USD

• Clearly overvalued price: 9.5–11.0 USD

• Current price classification: acceptable to hold

• Worth waiting for a better price: yes

• Target holding period: 3–5 years

• Expected annualized return: conservative -1% to 0%; base 5% to 7%; bull 15% to 18%

• Maximum loss risk: around 50%; triggered by several consecutive quarters of services weaker than expected, software net retention falling below 103%, and valuation multiples compressing into the traditional services-company range

• Signals that trigger reassessment: Software net retention below 103% for two consecutive quarters

• Services bookings below -5% year over year for two consecutive quarters

• Adjusted EBITDA margin below 29% for two consecutive quarters

• Arsenal files a new large selldown disclosure

• After H2 2026, the company still cannot pull core growth out of the 0%–4% range

【Ideal/Fair Buy Price】3.6–4.1 USD Rationale: This provides roughly a 20% margin of safety versus conservative intrinsic value of 5.1 USD, and this range already incorporates the risk that services recover more slowly than expected and valuation remains discounted.

【Valuation Range】

• current: 5.27 (as of the 2026-06-11 close)

• bear (conservative · ideal buy zone): [3.6, 4.1]

• base (reasonable · acceptable hold zone): [5.2, 7.0]

• bull (optimistic · above clearly overvalued line): [9.5, 11.0]

Key Data Table

This table compresses the most important data in the report: the problem is growth, not cash flow; the problem is services, not software; and the governance variable is the exit of the large shareholder and whether the new CEO can reshape the organization correctly, not financial fraud.

Research Uncertainties

• The company does not disclose an exact split between maintenance capex and growth capex, so owner earnings can only be estimated as a range and cannot reach very high precision.

• This one-off study did not rebuild a complete daily valuation percentile sequence since listing, so I provide only a directional judgment on historical valuation position, not a mechanical percentile.

• Whether Arsenal has begun reducing its stake after 2026-04-14 was not visible in updated public ownership-change filings as of the base date; this affects the near-term stock price rather than the core business.

• Management's view that services will recover in H2 2026 is currently more forward-looking language and still needs at least two quarters of data validation.

• AI-related products are still mostly at the stage of feature releases and organizational investment. Standalone revenue contribution disclosure is insufficient, making the true incremental impact hard to quantify.

References

• Certara 2025 annual report, Q1 2026 report, 2026 proxy filing, Q4 2025 and Q1 2026 earnings releases and conference calls.

• Certara announcements on CEO transition, sale of the regulatory-writing and medical-writing business, Arsenal investment, and director-nomination arrangements.

• FDA and EMA official documents and program pages on MIDD and PBPK.

• Latest public materials from comparable companies: Simulations Plus, Schrödinger, IQVIA, ICON.

• Current stock price, market cap, and risk-free-rate references: financial quotation tools, market quotation pages, and official U.S. Treasury/FRED data.

Other Tickers Mentioned in the Report

• SLP.US — The listed comparable closest to Certara's core modeling-software character, used to compare a cleaner software structure and higher valuation.

• SDGR.US — A physics-based simulation and AI drug-discovery platform, used to compare another path with a stronger technology narrative and weaker current profitability.

• IQV.US — The new CEO's former employer and a scale reference for life-sciences data and services platforms.

• ICLR.US — A traditional full-service CRO reference, used to compare governance and valuation discounts in services businesses.

• RXRX.US — A common comparison in the AI drug-discovery context, used to highlight market pricing differences around the "AI pharma" narrative.

• NVS.US — The collaboration background with Schrödinger is often used by the market to illustrate the commercialization value of front-end computational platforms.

• LLY.US — Repeatedly mentioned by Schrödinger through the Ajax transaction, reflecting a path through which technology platforms are validated by large pharma.

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