Key Takeaways

  • Software-only biology platforms struggle to survive because venture investors eventually force them to pivot into internal drug discovery pipelines to prove financial returns.
  • Fabian Theis observed that startups begin with broad virtual cell architectures, but end up diverting capital to advance one or two drug assets discovered along the way.
  • Defensible computational biology businesses build proprietary wet-lab data engines, such as organoid perturbation atlases, rather than standalone algorithm suites.
  • European biotech hubs like Germany offer early non-dilutive capital, but lose companies to the United States when later-stage venture capital is required to scale wet-lab operations.

The Platform Trap in Computational Biology

Software founders often assume that building a foundation model of a cell creates an obvious business. Train on public datasets, generate synthetic biological predictions, and license the platform to big pharma.

Fabian Theis, Director of the Computational Health Center at Helmholtz Munich, watches this playbook fail repeatedly. While tools like scConcept, Nicheformer, and CellFlow advance biological modeling, selling pure computation rarely sustains a venture-backed company.

“And I've been seeing this again and again that these platform type of ideas are super attractive,” (play from 38:48) Theis noted. “But turning this actually into a monetary thing is not as obvious.” (play from 38:59)

Pharma executives hesitate to write large software license checks for untested digital abstractions. Meanwhile, venture capital investors look at the balance sheet and demand tangible enterprise value. Because the software alone does not capture the upside of an approved therapy, investors steer the company toward internal therapeutics.

“So, I saw so many of these platform companies pivoting at some point to we make a drug,” (play from 39:16) Theis explained. “And yeah, at the side, of course, we have our virtual cell, whatever business, but you know, we have two drugs that we found and now we put a lot of money in.” (play from 39:22)

Once a startup funds wet-lab validation and clinical development for two molecules, it ceases to be a machine learning company. It becomes an early-stage biotech with a high-burn computational division.

Pair Algorithms with Wet-Lab Engines

Founders who want to build durable platforms must stop treating wet-lab biology as an afterthought. Pure algorithms cannot build a moat because public models and academic baselines catch up quickly. The real defensive asset is a closed-loop system where automated wet labs generate perturbation data that feeds algorithmic predictions.

“I think if you couple it with data generation not just with the modeling part, I think this is where sweet spot hits where then also actually pharma gets buying,” (play from 41:20) Theis said.

When startups construct targeted perturbation engines, like organoid perturbation atlases or active learning setups like LabCompass, the data asset becomes impossible for pharma incumbents to replicate internally. That proprietary biological output provides the commercial leverage that algorithms alone cannot deliver.

“More of the companies I see, they build their moat not just around a virtual cell model but in combination with specific data,” (play from 41:27) Theis observed. “This is then of course more value creation.” (play from 41:36)

The Scaling Bottleneck Outside the US

Building an automated wet lab alongside a software platform demands heavy capital. For European founders, this reality creates a geographical divide. While initial research grants, university spinout programs, and non-dilutive seed money exist in regions like Germany, growth-stage venture capital remains concentrated in North America.

“The thing is where Europe and particular also Germany drops off is at the scaling part,” (play from 35:12) Theis said. “Early VC and non-diluted type of money is there. But then when you go into scaling, these big VCs that take these bigger risks are not closely as present.” (play from 35:16)

If you build a hybrid lab-and-code platform, you will inevitably need large-check venture firms to finance the biological infrastructure before enterprise licensing revenue arrives.

What to Do With This

Audit your product roadmap tomorrow morning. If your company relies purely on training public biological data or public chemical libraries, map the exact proprietary experiment you must run in a wet lab to validate your next three predictions. Secure access to a contract research organization or automated bench partner this month, and frame your commercial pitch to pharma around that proprietary dataset rather than your model parameters.