Build vs. Partner: The FAL Decision
This is the most critical strategic fork in the road. The Ginkgo story is instructive: they burned $6 billion and lost revenue for three straight years trying to build and own the full FAL stack. They are now pivoting to software + orchestration (Cloud Lab). Recursion built automation but their real moat is the 50-petabyte dataset, not the robots. The hardware is being commoditized.
Do NOT build your own FAL in Phase 1 or 2. Be the intelligent orchestration and AI layer. Own the algorithms, the data, and the customer relationships. Partner or access-on-demand for the robotic execution layer. Revisit FAL ownership only at scale, when $50M+ ARR justifies the CapEx.
Why NOT building saves you
- CapEx avoided: A world-class FAL costs $50–200M to build and staff
- Ginkgo's cautionary tale: $6B losses, decommissioning benches in 2026
- Hardware commoditizing: Opentrons, Hamilton, Tecan, Automata all offer modular APIs
- Faster time-to-customer: Partner CROs exist — Syngene, Aragen, Lambda, GVK Bio
- India CRO base: $3B+ CRO market, underutilized for AI-driven workflows
- Margin clarity: Software EBITDA margins 70%+; hardware-heavy biotech 20–40%
The 3-Phase FAL Evolution
Phase 1 (0–24 months): Partner CRO Model
Use Syngene, Aragen, GVK Bio as execution partners. LabOS orchestrates their instruments. You own the AI layer and data.
Phase 2 (24–48 months): Hybrid Modular Pods
Deploy 3–5 Automata/Opentrons modular pods co-located with CRO partners. "FAL Pods" as a branded sub-service. Modular, incremental capital outlay.
Phase 3 (48–72 months): Own FAL
Build India's first purpose-built Fully Autonomous Lab in Hyderabad/Bengaluru Genome Valley. License LabOS to others. Infrastructure play.
Ideal Partner Ecosystem (India)
| Partner Type | Key Names | What They Provide | Our Leverage |
|---|---|---|---|
| CRO (Wet Lab) | Syngene, Aragen, GVK Bio, Lambda | Lab execution, instruments, scientists | AI-driven protocols, data ownership |
| Cloud Lab | Ginkgo Cloud Lab, Arctoris | Remote robotic execution, global access | Orchestration layer + India IP translation |
| Genomics | MedGenome, Strand Life Sciences, 4baseCare | India genomic cohorts, NGS infrastructure | AI interpretation, clinical integration |
| Hardware | Automata, Opentrons, Hamilton | Modular robots, APIs | Workflow orchestration, data capture |
| Cloud | AWS Hyderabad, Azure India, Google Cloud | HPC, storage, AI infrastructure | Negotiated credits, co-sell motions |
CapEx comparison
| Approach | Upfront CapEx | Time to first customer | EBITDA margin at scale |
|---|---|---|---|
| Own FAL (Ginkgo path) | $50–200M | 36–48 months | 20–40% |
| Partner CRO (our Phase 1) | <$2M platform | 6–12 months | 55–70% |
| FAL Pods (Phase 2) | $5–15M modular | 24 months | 45–60% |
APAC execution overflow routes through Singapore BioHelix and MiRXES i4.0 infrastructure in Phase 2+.