How one company is using machine learning and robots to unlock the power of biology
We started Zymergen a few years ago with an ambitious mission: to unlock the power of biology. We believed that if we did it right, it would be a catalyst for the next industrial revolution. Sectors traditionally associated with biology — like healthcare and agriculture — are just the beginning. New materials, innovative products, and solutions to some of the world’s biggest challenges are possible. But of course, we knew we would only accomplish these things if we could engineer biology at a level of predictability, reliability, and performance previously unattainable through traditional approaches to R&D.
To date, the practice of biology has been an artisanal craft reliant largely on individual genius. This works at certain scales and in some industries, but it simply won’t enable us to make biology industrially relevant across major industries.
“We believed that if we did it right, it would be a catalyst for the next industrial revolution.”
That’s why we’ve been focused on finding a new way to harness the potential of biology at Zymergen. One that frees us from both the constraints of human intuition and the inherent error of manual lab work. And so, we’ve built a general-purpose, data-driven platform for designing and optimizing biological systems. This platform amplifies the work of biologists, circumventing the need to rely on a single hypothesis. Essentially, we’ve built a search function for biology.
One way of understanding this is by analogy with the early days of the web. In the mid-1990s, Yahoo used curators — librarians and other domain experts — to find, categorize and rank web pages. This was a human-driven approach, much like the scientists who use pipettes to move small amounts of liquid around. It worked well when the web was small, but as the web grew, that human-curated approach quickly lost power because it was not scalable. Ultimately this approach was replaced by Google’s PageRank, a simple algorithm that counted links between web pages as votes, to rank content. Over time, Google has continuously improved the quality of its algorithms and today is uniquely able to understand and surface what is important on the web. What we’ve created is an engineering approach to biology that searches genetic space in the same way Google optimizes searching the web.
But successfully engineering biology is difficult because:
- The links between genes and traits are very poorly understood. We don’t know what most genes actually do and we understand little about cellular metabolism — the enormous number of interlocking and non-linear chemical reactions that take place inside the cell. Given this complexity, human ability to understand and predict what is happening is really limited.
- The design space is enormous. A conservative estimate of the design space for microbes — basically all the combinations of all the base pairs in a microbe’s DNA — is 4³⁰⁰⁰⁰⁰⁰. By way of comparison, there are only 10⁸¹ atoms in the universe and AlphaGo, Google’s recent Go playing software, searched a game space of only 10³⁶⁰. These numbers are so large it’s almost impossible to get your head around them. Simply put, the design space is too complex for us to comprehend and too vast to search manually.
- Traditional R&D processes are prone to error. Benchwork in molecular biology is a fiddly thing. It’s prone to error in part because of the precision required to perform many procedures. No one who’s ever done bench work was all that surprised that Amgen could recreate the findings of only 6 of 53 ‘landmark’ preclinical cancer studies.
As Zach likes to say “What we don’t know about biology vastly exceeds what we do.”
In just over two years, we’ve built a platform that vastly increases the speed of discovery. We do this in a way unlike any other company. We use our platform to systematically analyze everything first while other companies use their technology to execute human ideas. Our way is faster, more efficient, and far more precise.
Today, we’re excited to share that we’ve raised $130M in Series B funding led by SoftBank Group. SoftBank understands our business and our technology that helps Fortune 500 companies employ microbes to make better products and materials; they’re known for identifying market leaders, which makes it all the more exciting that they’ve identified us. We’re also excited our prior investors, including Data Collective (DCVC), True Ventures and others continue to demonstrate their confidence in our vision by returning to join the new round. To us, this reaffirms that investors recognize the transformative power of our technology as well as the potential of our business.
“What we don’t know about biology vastly exceeds what we do.”
We’re also thrilled to share that there will be two additions to our Board: former U.S. Secretary of Energy and Nobel laureate Dr. Steven Chu; and SoftBank Group International Managing Director Deep Nishar, previously a longtime senior executive at Google and then LinkedIn. As we think about building our business for the long term, governance matters a lot to us. Steve and Deep have extensive and diverse expertise that will be invaluable to Zymergen as we scale. These are two bright minds in science and in business, and we’re honored they share our vision.
Now we get to roll up our sleeves and put this funding to work. Zymergen has proven technology and has brought economic value to customers, and this new investment will help us increase our growth rate. We are building a 100-year-company because we know that while the introduction rates of new materials take longer in this field than others (~5–10 years), their product-life is much longer as well (~50–70 years) — so our impact can literally last a century.
If you share our excitement and vision, we hope you’ll think about joining us on this journey.
-Josh Hoffman, Zach Serber & Jed Dean