Schrödinger at KeyBanc Forum: Strategic Shift to Hosted Contracts

Schrödinger plans to convert 75% of its software contracts to hosted models within three years; software revenue was $200M in 2025 with ACV of $198M and management guides 10%-15% ACV growth for 2026. The hosted transition is expected to be cash-flow neutral but will depress revenue recognition in 2026 as bookings become ratable; the company targets adjusted EBITDA profitability by 2028. New product Predictive Tox and AI integrations (partnering with LLM providers such as Anthropic) are cited as key growth drivers, and management remains open to M&A to augment the platform.

Analysis

The company’s conscious pivot toward running and monetizing a continuous, compute-heavy platform creates a second-order moat: by operating customers’ workflows you accumulate structured simulation outputs that are uniquely useful to train and validate large models, which in turn increases switching costs and creates optionality for higher-margin, usage-based products. That same dynamic also reallocates economic value toward hyperscaler partners — expect a meaningful rise in third-party cloud spend per incremental user that will show up as variable margin pressure before higher ARPU emerges.

Headline revenue volatility from timing and accounting will be misunderstood by many investors as demand weakness; in practice the market should re-rate once recurring consumption metrics become visible. The clearest short-term read-through will be product-level usage and renewal cadence at top-10 customers — those datapoints will drive a rapid multiple expansion or contraction within 2–6 quarters depending on conversion speed.

Competitive dynamics favor firms that combine physics-forward simulation with data ops: pure-ML vendors struggle to explain failure modes on novel chemistry, while incumbents that control both model and compute can monetize model-output datasets. Conversely, relinquishing some internal program execution reduces a proprietary feedback loop; if that feedback loss cannot be fully replaced by customer-derived signals, adoption and clinical translation could slow materially over a multi-year horizon.

Key catalysts to watch are: (1) material uptick in platform usage from at least one tier-1 pharma, (2) proof points where early in-silico signals prevent costly downstream failures, and (3) any commercial agreements that shift compute economics. Tail risks include geopolitically driven data-residency constraints and adverse changes in large-model access/pricing that would compress margin on compute-intensive workflows.

Key Decisions for Investors

• Long SDGR equity (size 2–4% portfolio) on a disciplined pullback; horizon 12–36 months. Rationale: capture multiple expansion when recurring usage metrics and enterprise renewals reappear; risk: near-term headline revenue noise and execution; set a 20% trailing stop or hedge with puts if downside protection is required.
• Buy SDGR 18–30 month LEAPS calls (allocate 1–2% notional) to gain convex upside to successful platform/ML integrations. Reward: asymmetric upside if agent-like automation broadens user base; risk: premium decay and delayed adoption — cap position size accordingly.
• Pair trade (directional): long SDGR / short IQV (or a large CRO proxy) sized to neutralize market beta, horizon 9–18 months. Thesis: software-driven, usage-based economics should outpace legacy clinical services as budgets move upstream; risk: CROs benefit if outsourced trial activity accelerates.
• Tactical: overweight large-cap cloud infra (AMZN or MSFT) for 6–12 months to play increased compute demand from simulation-first workflows. Risk/reward: modest beta to secular cloud growth with lower idiosyncratic execution risk than single-name software.