Pathos raised a $365M Series D in May 2025 at approximately $1.6B post-money valuation.

Before that, Pathos closed a $62M Series C in October 2024 led by NEA at a $600M post-money valuation, bringing total funding at that point to $102M.

Earlier rounds included participation from Revolution Growth, Lightbank, Builders VC, and Tempus.

Across its known rounds, Pathos has raised approximately $467M in total primary equity financing.

Pathos is an AI-native oncology development company built around PathOS, a proprietary platform for sourcing, selecting, and advancing cancer drug programs.

PathOS has three connected components. Scout scans investigational oncology therapies, maps each to the patient subgroup most likely to respond, and ranks programs for pursuit. Sprint moves selected assets through development milestones using small, semi-autonomous "Sprint Pods," each responsible for taking one asset from one clinical inflection point to the next. Foundry is the shared intelligence layer underneath both, an oncology foundation model that learns from each program and ties model-generated insights to a wet-lab validation loop so biological hypotheses can be tested rather than remain purely computational outputs.

The platform is trained on what Pathos describes as more than 200 petabytes of multimodal oncology data linked to patient outcomes, roughly 50 times the size of TCGA, spanning genomics, transcriptomics, imaging, pathology, longitudinal clinical records, and treatment histories.

In practice, Foundry ingests conference proceedings, regulatory filings, published trial data, and proprietary real-world evidence. Scout then identifies assets for potential acquisition or development. The clearest public example is the May 2026 acquisition of a majority stake in DeuterOncology to advance the MET inhibitor DO-2. After an asset enters a Sprint Pod, the team uses the platform to design biomarker-driven studies, simulate trial scenarios, define enriched patient populations, and make go/no-go decisions faster than a conventional development hierarchy.

Pathos's current clinical pipeline includes pocenbrodib, a CBP/p300 inhibitor being developed across prostate, breast, and multiple myeloma, and P-500, a brain-penetrant PRMT5 inhibitor for advanced solid tumors including high-grade glioma and uveal melanoma. DO-2, the third-generation MET kinase inhibitor sourced through Foundry, is a third program and the first asset Pathos has publicly described as identified, evaluated, and acquired entirely through the platform.

Pathos operates as a hybrid AI-biotech platform spanning oncology data and model infrastructure, clinical-stage asset development, and biopharma partnerships.

Its monetization is not seat-based software. Value is captured through collaboration and platform economics with pharma partners, through ownership or majority stakes in clinical-stage assets that appreciate as they hit development milestones, and through downstream therapeutic upside from programs the platform helps select and de-risk.

The Tempus and AstraZeneca foundation-model collaboration illustrates the platform layer. In that three-party arrangement, Pathos serves as development orchestrator and model builder, committing capital to data and compute while receiving compensation for its development role. This makes Pathos both a buyer of infrastructure and a provider of execution intelligence within the same structure.

The Sprint Pod model drives the asset economics. Each pod is intended to move one program from one value-inflection milestone to the next, where biotech valuations typically reset. The DeuterOncology majority-stake acquisition shows the platform being used for corporate development as well as trial optimization: Foundry identified and evaluated DO-2 as one of four major portfolio decisions made in early 2026.

The cost structure is heavier than a pure software business. Data licensing, cloud compute, AI research and engineering, wet-lab validation, and clinical operations run at the same time. The model depends on amortizing one expensive AI and data core across many programs, so each new asset does not require building a new analytic system from scratch.

Pathos competes in a market where serious rivals are also moving toward vertical integration across data, models, trial design, and asset economics. The competitive question is less who has AI, and more who controls proprietary patient data, workflow distribution, and downstream drug economics.

Tempus is Pathos's most important strategic partner and its clearest potential competitive threat. Tempus offers multimodal real-world data, foundation models, agentic tools, clinical trial matching, biological modeling with organoids, and trial-network capabilities. It also expanded through acquisitions, acquiring Deep 6 AI in March 2025 for real-time EMR-based patient identification and announcing a deal to acquire Personalis in July 2026 for longitudinal monitoring and MRD capabilities.

The Pathos-Tempus relationship is structurally double-edged. Pathos gains data scale it could not assemble independently in the near term, but a meaningful part of its foundation-model strategy depends on infrastructure and data access controlled by a company with overlapping commercial ambitions. If Tempus moves further upstream from enablement into economic participation in drug programs, the distinctiveness of Pathos's position narrows.

Flatiron and Foundation Medicine, within the Roche orbit, represent a deep incumbent ecosystem in oncology evidence generation. Foundation Medicine brings broad genomic profiling and biopharma services, while Flatiron contributes oncology EHR-derived real-world data, global RWE capabilities, and trial design support. This stack is less AI-forward in branding than Pathos, but it is deeply embedded in pharma procurement and carries high trust for biomarker strategy and regulatory-facing evidence packages.

ConcertAI competes most directly in Pathos's trial-design lane, marketing agentic and generative AI for study design optimization, site selection, cohort generation, protocol refinement, and enrollment forecasting on top of a claimed 13-million-patient oncology dataset and the CancerLinQ network. The key difference is go-to-market: ConcertAI is a B2B SaaS and data vendor to life sciences sponsors, while Pathos is using similar capabilities to also originate and operate internal programs.

Owkin is the closest conceptual peer among AI-native oncology development companies, combining continuously enriched multimodal data, spatial and multimodal biomarkers for patient selection, wet-lab infrastructure, and external clinical networks. Owkin places more emphasis on federated data access where hospital partners resist central pooling, while Pathos is more tightly coupling AI trial design with internal asset incubation and acquisition.

IQVIA and Veeva remain deep systems of record in regulated clinical development. IQVIA owns the incumbent clinical data and services layer with deep trial design, CRO, and commercialization reach, while Veeva owns the regulated content, document management, and clinical workflow layer. Neither is AI-native in the way Pathos is, but both are sticky in sponsor procurement and can bundle AI capabilities on top of existing relationships faster than Pathos can displace them.

Caris Life Sciences combines comprehensive molecular profiling, multimodal datasets with real-world outcomes, trial matching, and a large alliance network. Its data flywheel is tied to clinical testing and established laboratory infrastructure, which can make its data fresher and more commercially actionable for biopharma sponsors that buy trial enablement from trusted testing vendors before they buy broader platform change.

Pathos's expansion logic runs in three directions: broadening the asset portfolio the platform can source and develop, deepening its role in pharma partner workflows, and extending its geographic footprint through the clinical infrastructure it is building in Europe. The TAM expansion case rests on whether Foundry can repeatedly source assets, support partner programs, and improve trial execution across markets.

The most immediate TAM expansion is repeating the DeuterOncology pattern at scale. Foundry is now used not just to optimize existing programs but to identify, evaluate, and acquire underappreciated Phase 1 and Phase 2 oncology assets with clear mechanistic logic and a biomarker-defined path to differentiated positioning.

Management described DO-2 as one of four major portfolio decisions made through Foundry in early 2026 alone, implying the platform is generating a pipeline of acquisition candidates rather than occasional one-off deals. If that sourcing cadence holds, Pathos's TAM expands from improving outcomes on a fixed set of partnered assets to owning a broader share of oncology asset economics across multiple mechanisms and tumor types.

Each molecule Pathos advances can also extend across multiple indications using the same multimodal subgrouping logic. pocenbrodib is already positioned across prostate, breast, and multiple myeloma, and P-500 is framed for combinations beyond its initial glioma and uveal melanoma settings.

Pathos's strategy includes building through biopharma partnerships as well as through wholly owned development. The AstraZeneca and Tempus collaboration indicates that Pathos can act as a platform and development orchestrator for large oncology sponsors while advancing its own assets.

That creates a second revenue layer: pharma companies that want multimodal modeling, biomarker-driven trial design, and patient selection intelligence, but do not want to sell or out-license their assets to Pathos, can still pay for platform access and co-development services. This expands Pathos's addressable market beyond asset ownership into fee-based and shared-development relationships.

FDA's Project Optimus and its continued emphasis on molecularly defined development paths reinforce demand for simulation-heavy, biomarker-aware trial design, especially for targeted agents and combinations where crude Phase 1 dose escalation is less sufficient.

The DeuterOncology acquisition gives Pathos an immediate European operating footprint instead of requiring it to build one from scratch. DO-2 has already completed Phase 1 dose escalation across sites in the Netherlands, Belgium, and France, providing an operating base for pan-European trial execution, regulatory engagement, and deal sourcing.

Europe's health-data infrastructure is becoming more formalized through the European Health Data Space regulation and the Clinical Trials Information System, both of which improve data portability and harmonize trial administration across member states over time. For a company whose core advantage is multimodal data linked to outcomes, better cross-border data access and more efficient multinational enrollment infrastructure are structural tailwinds, especially for biomarker-defined patient populations that are too small for single-country development.

Data dependence: Pathos's foundation-model strategy depends on Tempus's data, cloud infrastructure, and commercial terms under a multi-year master agreement disclosed in Tempus's own SEC filings, so any tightening of access terms, field restrictions, or strategic misalignment between the two companies, which share a founder in Eric Lefkofsky, could weaken the data advantage behind Foundry while Tempus expands into overlapping commercial territory.

Proof burden: Unlike a software company where product-market fit can be measured through retention and expansion metrics, Pathos's business model, platform credibility, partnership economics, and asset valuation depend on whether its integrated AI and data architecture delivers measurably better clinical outcomes than conventional biotech decision-making, a standard that requires successful trials rather than model benchmarks.

Principal conflict: As Pathos accumulates majority stakes in clinical-stage assets and uses Foundry to source acquisitions, pharma partners considering co-development arrangements may view sharing program data and biological insights with a company that is also building a competing proprietary pipeline as an unacceptable risk, which could limit the breadth of biopharma partnerships available to Pathos as its asset portfolio grows.