Packed in dry ice and arriving in an oversized shoe box, the glass vials that clients ship to Zymergen Inc. don’t look like much.
What’s inside those vials—designer microbes that can cost $500,000 each to genetically engineer—are the biological building blocks used to make materials in everything from medicine and flexible electronics to radar resistant paint for stealth bombers.
Zymergen’s 200 employees and army of robots take it a step further. The start-up uses machine learning to instruct its robots how to genetically engineer microbes. Those microbes are then used to grow new and existing materials more efficiently.
Venture investors love the idea, with SoftBank Group Corp. now leading a $130 million funding round in Zymergen to bankroll its expansion, and former US secretary of energy and Nobel Laureate Steven Chu joining the board.
“Once you have something that’s radically new and you can train those microbes to make high volumes of the stuff then you really are in a different world,” Chu said in an interview. Zymergen is unusual because it can reliably reproduce results and operate a platform for any microbe rather than just one or two, which has been the industry standard, he said. “We are just beginning to scratch the surface on all of this.”
Along with Ginkgo Bioworks Inc., Zymergen is one of a growing number of AI-powered biotech start-ups to recently secure cash and confidence from investors to engineer microbes.
The Emeryville, California-based company will use the infusion to buy more robots for its labs—it has dozens now—and find more customers through its unusual sales process.
“Outbound engagement is an invasive scientific back and forth - it’s not enterprise software which you can (sell) on the golf course,” said Zymergen board member and early investor Rohit Sharma of True Ventures.
Since launching three years ago, Zymergen has signed what chief executive officer Joshua Hoffman says is “a handful” of Fortune 500 customers. He declined to share the identities of those customers, citing their desire to preserve a competitive advantage.
The work flow goes a little something like this. Zymergen tests a customer’s microbe and determines how efficiently it can convert raw material to the polymer, pesticide or other desired substance. Zymergen’s algorithms then suggest making 1,000 or so changes to the microbe’s genetic material to increase yield and cut costs. That’s when the robots take over, injecting the suggested DNA snippets into the specimens, testing their properties, collecting data about each new combination and feeding that information back into the data trove.
The system continues grinding through various combinations, until it can produce a new and better microbe than the one initially shipped for testing in the oversized shoebox. The newly designed microbe is then given back to the customer and can be used to mass manufacture items like clear flexible polymers for electronics.
“With startups, they say they can do new things all the time, but it’s usually at a higher cost. This is different,” said Sharma.
Zymergen charges a subscription fee and takes a share of the improved margins over a certain time period—the goal being to cover costs and get a slice of their customers’ upside. The company declined to share revenue.
While improving existing products is important, creating new microbes is the company’s long-term focus. Historically, discovery of new microbes has been cost and time prohibitive.
Hoffman says that Zymergen’s self-learning system slices the cost of creating new microbes from several hundred million dollars to tens of millions, and cuts the time frame from 10 or 15 years to just two or three.
George Church, professor of genetics at Harvard Medical School, says Zymergen’s approach is “nearly perfect, ” and said the company, along with Amyris Inc. and others, are providing a boon for the discovery of new drugs and materials.
“Most of these processes now in biotech and manufacturing are now very artisanal,” he said, adding that is starting to change. “The price for reading and writing DNA have come down more than a million fold over the past few years -- way more than electronics during the same period.”
While the company says it’s past the “vaporware” stage of a start-up and has real systems at work, Zymergen is operationally complex. The biggest challenge, according to Hoffman, is executing.
Mark Breidenbach, an analyst at Roth Capital Biotech Research, said the company’s complexity could be a stumbling block.
“It sounds like a lot of moving parts to me,” he said. “It may sound good to try and marry a lot of technologies, but I think it’s more geared toward attracting investment from people in the industry who are not experienced health-care investors.”
Softbank led the recent growth round with participation from Iconiq Capital, Prelude Ventures and Tao Capital Partners, bringing total outside funding to the three-year-old startup to $174 million. Existing investors, including Data Collective Venture Capital and True Ventures also contributed.
The company declined to disclose the valuation. Investors valued Zymergen at approximately $430 million after the round, according to Equidate, a stock market for private companies.
Along with Chu, SoftBank managing director Deep Nishar is joining the board. Bloomberg