Junction Bioscience

What do they actually do

Junction Bioscience is building an internal “hypothesis engine” that loops between molecular simulations and wet‑lab experiments to propose and refine candidate molecules, with an initial focus on neuroinflammation and immunology. There’s no public SaaS or API; current activity looks like an R&D platform used by the company’s scientists with select partners, not a finished commercial product yet (YC profile; company site; YC social post).

In practice, they start with a disease question, generate molecular hypotheses, run in‑silico scoring, test a short list in the lab, and feed results back to the models to guide the next experiments. Public signals point to a very small team and early partnerships (including a first six‑figure collaboration) rather than a broad customer base at this stage (YC profile; LinkedIn).

Who are their target customer(s)

• Large pharma discovery and preclinical R&D leaders: They run long, expensive design–test–learn cycles and need to reduce low‑value experiments to move programs toward clinical candidates faster (YC profile).
• Small biotech discovery teams with limited lab capacity: They can’t afford broad screening or large compute teams and need help deciding which molecules to make and test next to conserve cash and time (YC profile; company site).
• Academic and translational research groups: They have mechanistic questions but lack an integrated compute + wet‑lab loop to rapidly iterate and validate hypotheses; they need collaboration that produces publishable results and de‑risks follow‑on industry work (YC profile).
• CROs and core facilities executing assays for others: They lose time and money on low‑value or poorly prioritized experiments and want clearer, model‑driven experimental plans to increase throughput and reduce wasted assays (YC profile; company site).

How would they acquire their first 10, 50, and 100 customers

• First 10: Run high‑touch pilots with the earliest partners: extend the initial pharma collaboration, convert the academic case study into a sponsored research project, and add a few focused sprints with a mid‑sized biotech plus a core facility/CRO to generate measurable before/after reductions in experiments and timelines (YC profile). Keep scopes small, milestone‑based, and fast to contract so legal and procurement cycles don’t stall progress.
• First 50: Standardize “pilot packs” (fixed scope, timelines, deliverables), publish anonymized case studies, and run technical workshops/webinars to convert interest into scoped pilots. Build a lightweight CRO partner program, introduce legal/IP templates to cut cycle time, and hire a BD lead to systematize outreach and follow‑up at conferences and via investor/accelerator networks.
• First 100: Productize repeatable offers into service lines with clear SLAs and pricing, add reseller/white‑label deals with larger CROs, and offer flexible commercial models (fixed fee, milestones, pay‑for‑success). Support enterprise buyers with integrations, dashboards, training, and frame agreements while expanding distribution via foundations, international partners, and targeted roadshows.

What is the rough total addressable market

Top-down context:

Global pharma R&D spend is roughly $289B in 2024, and preclinical/non‑clinical work is commonly estimated at about 31% of total development costs, implying on the order of ~$90B annually is spent on discovery/preclinical activities that Junction aims to improve (Statista; CBO).

Bottom-up calculation:

Early deals look like mid‑six to low‑seven figures per partner per year; capturing just 0.01%–0.1% of the ~$90B discovery/preclinical pool implies ~$9M–$90M in annual revenue, i.e., tens of partner programs at $100k–$1M each if the platform demonstrates reliable decision quality gains (YC profile).

Assumptions:

• Preclinical share of R&D (~31%) is a reasonable proxy for the discovery/preclinical budget addressable by tools and partnerships.
• Customers will reallocate a portion of in‑house and outsourced discovery budgets to external hypothesis‑engine work if it reduces experiments and time to decisions.
• Deal sizes in the $100k–$1M/year range are achievable for scoped discovery programs combining compute and targeted wet‑lab work.

Who are some of their notable competitors

• Recursion: AI‑driven drug discovery company combining large‑scale biological data generation with in‑house compute and partnerships; competes for lab‑in‑the‑loop discovery programs.
• Insitro: Machine learning + high‑throughput biology platform pursuing partnered and internal programs; overlaps on data‑driven hypothesis generation and validation.
• Exscientia: AI‑enabled drug design company with a mix of partnerships and internal pipeline; focuses on improving design–make–test cycles.
• Generate:Biomedicines: Uses generative models to design proteins and validates them experimentally; adjacent in combining computational design with wet‑lab iteration.
• Atomwise: Structure‑based discovery using AI for hit identification and optimization via partnerships; addresses early decision‑making in discovery.