Recursion pricing

About

Recursion (NASDAQ: RXRX, headquartered in Salt Lake City, Utah) is an AI-enabled drug-discovery company — a “TechBio” platform that uses machine learning over enormous proprietary biological and chemical datasets to find and advance new medicines. Its core asset is the Recursion Operating System (Recursion OS), built on a dataset the company describes as over 50 petabytes spanning phenomics, transcriptomics, proteomics, ADME, and de-identified patient data. Recursion runs that platform on BioHive-2, an NVIDIA DGX SuperPOD with 504 H100 GPUs — one of the most powerful supercomputers wholly owned and operated by any pharma company.

Crucially, Recursion is not a software vendor with a price list. It makes money three ways: (1) pharma partnerships — multi-year discovery and development collaborations with companies like Roche/Genentech, Sanofi, Bayer, and Merck KGaA, structured as upfront cash + research funding + per-program milestone payments + tiered royalties; (2) an internal pipeline of owned drug assets in oncology and rare disease, whose value is realized through clinical progress, out-licensing, or M&A rather than a “price”; and (3) a platform/compute layer, anchored by a $50M NVIDIA investment (2023) and foundation-model distribution. Recursion IPO’d in April 2021 ($436M+ raised) and merged with Exscientia in an all-stock deal ($688M enterprise value) completed November 2024.

For the most current information, visit Recursion. Note: there is no public pricing page — Recursion’s monetization is deal-based and sales-led (partnering contact: partner@recursion.com).

Pricing summary : How Recursion’s pricing model works

Recursion’s “pricing” is pharma partnership economics, not subscription SaaS. A typical discovery partnership follows a waterfall: the partner pays an upfront sum (cash, sometimes plus an equity investment), funds the research as programs run, then pays per-program milestone payments as each candidate hits development, regulatory, and sales targets, and finally pays tiered royalties on net sales once a drug reaches the market. Value is metered by scientific and clinical deliverables — a program advancing a stage, or a delivered Phenomap — rather than tokens, seats, or API calls.

The headline numbers are large and flow toward Recursion. Roche/Genentech (Dec 2021): $150M upfront, up to 40 programs each with more than $300M of milestone potential, plus tiered royalties and Phenomap option fees that could exceed $250M. Sanofi: $100M upfront, up to 15 programs, up to $5.2 billion in aggregate milestones, royalties high-single-digits to mid-teens. Bayer: up to $1.5 billion in success-based payments across up to 7 oncology programs. Merck KGaA (2023): $20M upfront, up to $674M in milestones across 3 programs. Across all deals the milestone potential exceeds $20 billion before royalties — and by Q3 2025 Recursion had received over $500M in cumulative upfront and milestone payments, with ~$785M cash on hand.

What makes this different: in a corpus full of token-metered and per-seat AI tools, Recursion is the rare entry whose “customers” are pharma giants signing nine-to-ten-figure collaborations, and whose value metric is a drug program reaching a milestone. The platform (Recursion OS) and the compute (BioHive-2) are not sold by the unit — they are the moat that makes a partner willing to pay for a shot at a successful medicine. Revenue is recognized as milestones are earned, making it among the most outcome-based monetization models you’ll find anywhere.

Pricing by product

Sales motions across products: sales-led / business-development only — there is no self-serve product, no rate card, and no monthly plan. Every dollar is negotiated as a multi-year pharma collaboration or a strategic technology deal. Recursion’s IPO and partnership inflows fund an internal pipeline whose upside it keeps for itself.

Hidden costs : What Recursion partners actually pay

Because this is a partnership model, “hidden costs” work differently than in SaaS — the real economics are about how much of a deal’s headline value ever gets paid and when. A deal “worth up to $5.2 billion” (Sanofi) or “$1.5 billion” (Bayer) is a success-based ceiling: those totals are only reached if programs survive a brutally low-probability drug-development gauntlet. Most milestone potential is contingent and back-loaded — paid only as candidates clear preclinical, clinical, regulatory, and sales hurdles, each of which most programs never reach.

The other “cost” to watch is R&D burn against a finite runway: Recursion is a clinical-stage company spending heavily to advance both partnered and owned programs, with ~$785M cash reported and runway into 2027. For a partner, the budgeting reality is that milestone payments are spread over years and gated by results; for an investor, the question is whether realized payments (the >$500M) grow fast enough against the multi-billion-dollar potential.

Want to model partnership economics rather than a SaaS bill? Use the Recursion pricing calculator to sketch upfront + milestone + royalty scenarios, or see our outcome-based pricing guide for how these structures work.

Pricing evolution : Recursion pricing history and changes

Cadence

Tracked range: 2020–present. There is no public price list to chart with Wayback snapshots — the “evolution” here is the partnership book and platform, not a posted rate card. Figures are from Recursion IR press releases and SEC filings.

Notable changes

• 2021-04 — IPO on NASDAQ (RXRX) at $18.00/share, ~$436M+ raised — funds a partnership-plus-pipeline model.
• 2021-12 — Roche/Genentech collaboration: $150M upfront, up to 40 programs at $300M+ milestone potential each.
• 2023-07 — NVIDIA invests $50M and partners on foundation models — adds a platform/compute monetization layer.
• 2024-05 — BioHive-2 supercomputer (504 H100 GPUs) completed — the compute moat behind partner value.
• 2024-11 — Exscientia merger (~$688M, all-stock) completes — consolidates two AI-drug-discovery platforms.
• 2025 — Crosses $500M+ cumulative upfront + milestone payments (incl. a $30M Roche/Genentech microglial-map milestone); ~$785M cash, runway into 2027.

What’s unique : Recursion’s distinctive pricing mechanics

1. The customer pays you (a lot), in stages. Unlike every SaaS entry in this corpus, Recursion’s “buyers” are pharma giants paying upfront cash + research funding + per-program milestones + royalties — deals worth up to billions each. Revenue is earned as science progresses, not charged per unit, making it genuinely outcome-based.

2. The value metric is a drug program, not a token. Payment triggers are program advancement and delivered Phenomaps — scientific deliverables. A single program clearing a milestone can release tens of millions of dollars (e.g. the $30M Roche/Genentech microglial-map milestone), and a program reaching market unlocks tiered royalties for years.

3. Compute and data are the moat, not the SKU. Recursion doesn’t sell BioHive-2 GPU-hours or charge per petabyte of its 50PB dataset. The $50M NVIDIA stake, the 504-GPU supercomputer, and data deals (Tempus, Helix, Google Cloud) exist to make the platform valuable enough that pharma will sign milestone-and-royalty deals — monetization is one level removed from the infrastructure.

Strengths & weaknesses

Billing UX : Recursion billing controls and transparency

• “Billing” controls — There is no in-app billing. Economics are governed by multi-year collaboration contracts: an upfront payment, defined research funding, a schedule of milestone payments tied to specific program and regulatory events, and a royalty rate on eventual net sales. Scope changes go through contract amendments and program-governance committees, not a dashboard.
• Usage visibility — As a public company (NASDAQ: RXRX), Recursion discloses partnership economics, milestones achieved, cash position, and runway in SEC filings and quarterly earnings — arguably more transparency than any SaaS rate card, just at the deal level. The /partners page lists collaborators; per-deal upfronts and milestone ceilings appear in press releases and 10-K/10-Q filings.
• Payment options — Partner payments are invoiced milestone-by-milestone under negotiated contracts; technology partnerships (NVIDIA) can include equity investment (PIPE) alongside collaboration terms. There is no card checkout, no monthly plan, and no self-serve tier.

Strategic wins : Why Recursion’s pricing decisions worked

1. Outcome-based economics aligned with the buyer

By pricing on milestones and royalties, Recursion sells pharma exactly what it wants — shared risk on drug discovery — instead of asking it to pay per query for an AI tool. The structure turns a speculative platform into a partner willing to commit billions in potential value. See outcome-based pricing trends.

2. Diversifying the partnership book

Signing Roche/Genentech, Sanofi, Bayer, and Merck KGaA across neuroscience, oncology, immunology, and fibrosis spreads program risk and stacks upfront cash + milestone potential, so no single program’s failure sinks the model. Related: how AI companies structure pricing.

3. Building the moat that justifies the deals

The $50M NVIDIA investment, BioHive-2, the 50PB dataset, and the Exscientia merger all raise the platform’s credibility — which is what lets Recursion command large upfronts and milestone ceilings in the first place. See choosing the right usage metric.

Areas to improve : Gaps in Recursion’s pricing approach

1. Realized value lags the headline ceilings

Deals “worth up to $5.2 billion” read impressively, but >$500M received against $20B+ of potential shows how much value is still contingent. Tighter, clearer disclosure of near-term, high-probability milestones would help partners and investors model real cash, not just ceilings. See bill shock and cost unpredictability.

2. No platform-as-a-product revenue line

Recursion’s 50PB dataset and BioHive-2 are monetized only indirectly. A metered platform or foundation-model offering (the NVIDIA-cloud distribution hinted at in 2023) could add a recurring, predictable revenue line alongside lumpy milestone payments — diversifying away from drug-development risk.

3. Runway versus burn

The milestone model is back-loaded while R&D burn is front-loaded. Sustaining ~$785M cash and runway into 2027 depends on converting potential into realized payments faster — a structural tension every clinical-stage AI-pharma company must manage.

Key takeaways

1. Recursion has no price list — it monetizes via pharma partnerships. The model is upfront cash + research funding + per-program milestones + tiered royalties, sold deal-by-deal to companies like Roche/Genentech, Sanofi, Bayer, and Merck KGaA.
2. The value metric is a drug program reaching a milestone. A single deliverable (e.g. a delivered Phenomap) can release tens of millions; programs reaching market unlock royalties for years.
3. Headline deal values are ceilings, not cash. Up to $5.2B (Sanofi) and $1.5B (Bayer) are contingent; Recursion had realized over $500M cumulatively by Q3 2025 against $20B+ of total potential.
4. Compute and data are the moat, not the product. The $50M NVIDIA stake, BioHive-2’s 504 GPUs, and a 50PB dataset exist to make partners willing to pay — they aren’t sold by the unit.
5. Public-company structure shapes monetization. IPO (2021), NVIDIA investment (2023), and the Exscientia merger (2024) built the balance sheet and platform that underwrite the partnership model.

UBP implications

1. Outcome-based pricing reaches its purest form in pharma partnerships — value is metered by scientific and clinical milestones, the ultimate “pay for results” structure. Software vendors exploring outcome pricing can study how milestone waterfalls and royalty tiers align incentives. See usage-based pricing strategy.
2. Ceilings versus realized value is the core tension of any milestone model — practitioners should track received payments, not headline potential, when benchmarking outcome-based deals.
3. A platform can monetize indirectly — Recursion sells drug-discovery outcomes, not GPU-hours or dataset access, yet the infrastructure is what commands the deals. The lesson: sometimes the best “price” for a platform is the value it lets your customer create, captured via milestones and royalties.

Sources

• Recursion Partners page (live capture, accessed 2026-06-10)
• Recursion Platform page (live capture, accessed 2026-06-10)
• Recursion announces transformational collaboration with Roche and Genentech — $150M upfront, $300M+/program (accessed 2026-06-10)
• Recursion reports Q3 2025 results — >$500M cumulative, $30M Roche milestone, ~$785M cash (accessed 2026-06-10)
• Recursion announces $50M NVIDIA investment and foundation-model collaboration (accessed 2026-06-10)
• Recursion completes NVIDIA-powered BioHive-2 — 504 H100 GPUs (accessed 2026-06-10)
• Recursion–Exscientia merger — ~$688M all-stock, STAT News (accessed 2026-06-10)
• Recursion Pharmaceuticals IPO pricing/closing — NASDAQ: RXRX, ~$436M+ (accessed 2026-06-10)

Bottom line

Recursion (NASDAQ: RXRX) is an AI drug-discovery platform whose “pricing” is pharma partnership economics, not SaaS: upfront cash + research funding + per-program milestone payments (often $300M+ potential each) + tiered royalties on net sales, sold deal-by-deal to Roche/Genentech, Sanofi, Bayer, and Merck KGaA — alongside an internal owned-asset pipeline and an NVIDIA-powered platform (BioHive-2, 504 H100 GPUs; 50PB dataset). Aggregate milestone potential exceeds $20 billion before royalties, and Recursion had received over $500M cumulatively by Q3 2025. It IPO’d in 2021, took a $50M NVIDIA investment in 2023, and merged with Exscientia (~$688M) in 2024. There is no public price list — monetization is outcome-based and sales-led. Browse the pricing blueprint for more fully-researched company profiles.

Want to compare Recursion against other AI-platform companies? Browse the pricing blueprint.