Update: Pausing (Hibernating) MOLI Through the SynBio Winter

We founded MOLI (original known as Morf) with a singular mission: make industrial synbio viable. We define industrial synbio as the use of engineered biology for purposes beyond pharmaceuticals.

Our mission came with a clear vision: a world where synbio-made goods are cheaper, and thus market dominant, over fossil- and animal-based goods. Our vision is a world of sustainable abundance, with faster drug discovery and novel goods that harness the power of nature as collateral benefits.

When we first evaluated the industrial synbio landscape in 2023, we saw many exciting proof of concepts. We also saw huge problems to be solved on the technology infrastructure. We saw an opportunity to unlock viability for industrial synbio by introducing robotics, automation, AI, and first principles hardware design.

We learned about the status quo approach to infrastructure and pursuit of economic viability: harness economies of scale by building very large factories with very large, monolithic machines (bioreactors). We did not see many, if any, examples where this approach had worked, but it was the conventional wisdom. We saw clear issues:

  • Capital intensive & high risk — tens or hundreds of millions to build a large enough factory and unproven process viability until after the factory is built.

  • Slow — in building the factories and in scaling up a biological process to building-size bioreactors (i.e., takes years, often failing along the way).

  • Inflexible & failure prone — no ability to iterate on process, right-size production, or shift products; failed batches costing hundreds of thousands of dollars.

  • Inefficient — while economies of scale provides CapEx and labor efficiencies, it introduces inefficiencies on productivity, uptime, and utilities.

Further, we identified two emerging issues not yet being discussed:

  • No learning curve — like nuclear plants, cost of infrastructure had not improved over time.

  • No AI and automation — existing infrastructure had minimal data collection and automation.

Our core hypothesis was born: use robotics, automation, and first principles design to unlock a low-cost, modular infrastructure. If economically viable, the switch from monolithic to modular would eliminate the known and emerging problems with existing infrastructure, including unlocking a learning curve that would drive consistent cost reductions, as seen in industries like solar and batteries.

We set out to answer the first critical question: Could we build a sufficiently low cost and high performance reactor? As it turns out, yes. We combined novel innovations and existing best practices to build a 200L reactor for <$50,000, beating existing >$500,000 200L reactors on CapEx by 90%. We also validated reactor performance by using it to produce a nutraceutical ingredient, where we beat existing the yields and productivity reported in literature. We also designed the reactor with a robotics and automation foundation, and even made a mockup of a future modular factory (imagine rows of many moderately sized vats, with a robot on a track servicing each as needed). To show the AI potential, we built a Gen AI model that identifies bioproduction pathways for any given target molecule and evaluates the technoeconomics for each path, highlighting those that could be viably produced in our reactor infrastructure (and providing set points for automated deployment).

We built a low-cost, automated, AI-enabled, modular bioproduction system that flipped existing infrastructure shortfalls into benefits:

  • COGS efficiency — minimizes cost via productivity, uptime, labor, and utility efficiency gains.

  • CapEx & risk minimization — enables reaching viable COGS without having to build a large factory; validates economic viability before committing scale up capital.

  • Speed — eliminates the need to spend time scaling up to larger and larger tanks.

  • Flexibility — allows for constant product or process iteration and for distributing production near customers or feedstock.

  • Learning curve — our v2 prototype was drastically cheaper than our v1.

  • AI-unlock — collects huge amounts of data and unlocks AI-driven process optimization running across many small reactors.

We built an infrastructure that could not only unlock near-term scale up and cost reduction, but could amass valuable bio-data, which could be fed into AI models to develop ‘superintelligent bioproduction’: rapid, AI-driven development of processes to viably produce existing or novel molecules. We saw AI-driven development and autonomous infra as the future for industrial synbio.

We were ready to deploy our enabling technology and raise investment to continue developing it. Unfortunately, by the time we picked our heads up from months of tech development, most of the companies we set out to enable were gone or were struggling for survival. An industrial synbio winter had fully set in and dried up our potential customer base.

We explored and pitched becoming a producer ourselves (a Vertical Integrator, as Packy McCormick would say), but doing so was not possible given the available capital. It also would not have allowed us to continue progressing on the infrastructure.

Owning the infrastructure layer is incredibly valuable for a booming ecosystem. Look at AWS, Google, Apple, Microsoft, or Nvidia. But there have to be ‘developers’ to build on top of your infrastructure. We need healthy companies with promising industrial synbio processes. But there are currently not enough to make sense for us to continue. With no software developers, how could AWS have survived?

Thus, our update: we are pausing our efforts. We are putting MOLI into hibernation. We remain long-term industrial synbio believers. We know progress will continue. We believe that sooner rather than later the transformative potential of industrial synbio will eventually be realized. We know there will be advances and breakthroughs in the coming years. Eventually, we believe investment and vigor will return to the space. For now, though, we have to pause. But our technology exists and remains. We are happy to chat with any interested or curious companies, individuals, or investors. If we see enough potential users and investment develop, we can pick up where we left off.

To everyone who has supported this journey thus far, thank you. We hope to pick it up again in the future when the conditions are right.

Naya Kallison, CEO, naya@molibio.ai // Sean Rowan, CTO, sean@molibio.ai