The climb up the steel steps is dizzying—like ascending the tower of a European church, except the steps lead to a platform bolted to the side of a gleaming new chemical plant. Here in Brazil, under a brilliant blue sky, Eduardo Loosli, the plant manager, pauses to explain a vision of the future. "I used to manage a Molson Coors beer manufacturing plant, and it’s not all that different," he says, leaning on a railing and surveying the scene around us. Directly below is a cityscape of huge stainless-steel tanks. Out beyond the tanks, and stretching far into the distance, are dense greenfields of sugarcane.
Yeast turns grain into beer, Loosli says. Here, in this new plant, genetically engineered yeast created by the plant’s owners—the California biotech company known as Amyris—turns sugar into liquid fuel. At the end of the platform, Loosli points to two special "seed" tanks. "The yeast enters the system here," he says. When production starts, a glass flask of Amyris’s special strain will be poured into each tank, and the yeast will multiply until it becomes a thick, hungry broth. For two weeks, the yeasty stew chews up as much as 1.2 million liters of energy-rich cane syrup. The end product is farnesene, which can be adapted to a seemingly perfect replacement for petroleum-based diesel. Not only does farnesene-based diesel cut pollutants from vehicle exhaust pipes, but since it derives from cane syrup, it is also a renewable resource. These cane fields surrounding the plant are thus the rough equivalent to bottomless oil wells.
Amyris’s great innovation is deep inside the genetically modified yeast that chews up the Brazilian sugarcane. The yeast serves as a host for a set of DNA instructions—scientists call the organism a chassis, as if it were a simple platform, waiting for an engine. Depending on their goals, engineers at Amyris can outfit the yeast with a variety of genetic material that tells the yeast how to digest what it is fed. The result is a cell that can (at least in theory) ingest simple sugars and produce virtually anything. Indeed, if the yeast cells work as they’re supposed to, they promise not merely to change the energy industry by producing farnesene. They may also be programmed to transform the way many commodity materials are made. The first step would be petroleum-type materials. Rubber, chemicals, and medicines would follow.
At least that’s the idea. Already, in the short lifetime of the biofuels business, Amyris has become legendary—a stand-in for the sector’s breathtaking promise and now for its troubling descent. The company’s Brazilian plant is referred to as Paraiso, Portuguese for "paradise." It could be more aptly described as a grande esperanca, a great hope. Just a year ago, Amyris’s stock price soared to $33 a share. More recently, as the company reported $95 million in losses last quarter, it has plummeted to as low as $1.52. Meanwhile, a once-grand expansion plan has been scaled back. A plant at Sao Martinho, double the size of Paraiso, sits half-complete, vacant as of February. Amyris suspended production at another plant this year.
Everything now rides on success at Paraiso. This summer, as the final bolts were tightened and the last pipes sealed, Loosli readied himself to take command and find out what the plant can do. The future has become a matter of simple economics. If Amyris can produce farnesene efficiently here, the company will gain precious time to perfect its genetic technology. And if not? Then Amyris will likely capsize and pull an entire sector—an entire vision of the future—down with it.
A few weeks before I travel to Brazil to see the Paraiso plant, I discuss Amyris’s perilous situation with one of its founders, chief technology officer Neil Renninger. We meet at the company’s Emeryville, California, headquarters, in a small lounge decorated with black-and-white photos of Red Sox players and filled with comfortable leather couches. He reminisces about growing up in California—his dad had worked as an engineer at Intel, his mom as a schoolteacher. As he speaks, his slender fingers sometimes search for his iPhone, sometimes hang in the air, and sometimes touch tip-to-tip as he considers a thought.
Circles of exhaustion rim his eyes. His company is ailing and he wears it on his face. Just before my visit, Amyris had announced it had produced a million liters of farnesene in 2011, rather than the 6 million it promised. Its executives declared they would no longer make predictions about future production. "I’d be lying to you if I said that I didn’t look at the stock price," Renninger admits.
He always knew a startup is a gamble. As an undergraduate at MIT in the mid-1990s, before casino bouncers recognized MIT kids on sight, Renninger played on the institute’s notorious Black Jack Team. On nights and weekends, between engineering classes, Renninger traveled to Vegas and Mississippi River casino boats. He wadded up $100,000 of the team’s betting money in his pockets, and when a table ran hot, he’d clean up. The experience forged him. "The biggest thing I learned at MIT was go ahead and take risks because if you fail, you’ll land on your feet," he says.
He felt that way about Amyris. Ten years ago, Renninger was working in the lab of Berkeley chemical engineering professor Jay Keasling, a father figure in the field of synthetic biology. Keasling had come to believe that biology would ultimately follow the paths of engineering and computing, and that cells could in time be treated much like small factories, tiny machines whose insides behave like assembly lines. Keasling’s idea was that one day a biologist in front of a computer could piece together the virtual genes of a virtual organism, program and test it on a computer model, and then press print. From there, automated machines could produce the actual organism, which would behave exactly as the computer predicted. This is the vision that Renninger signed on to and what he spent years working toward. Today at Amyris’s California lab—downstairs from where Renninger is recounting the company’s history—a team of 40 scientists works on a computer program called Thumper. Essentially, the program allows scientists to rearrange a yeast’s genetic makeup and create new strains; more than 400,000 new strains are screened each week. In a sealed room, the fittest move from plates of colonies to half-liter fermenters—glass containers filled with soups of yeast and sugar, like mocha and cream. The fermenters resemble Cuisinart food processors. From there, the most promising strains are shipped to Brazil.
Back in 2002, Renninger recalls, before any of this was built, Keasling and his colleagues began sharing the details of their work with outside companies. One day, a postdoctoral student under Keasling named Vince Martin said, "We have some good technology and good people. Why don’t we do something with it?"
In response, Keasling invited Martin and three other postdocs—Renninger, Jack Newman, and Kinkead Reiling—to spin out a company. Their first goal: to genetically engineer a yeast cell to produce a chemical called artemisinic acid. Only one plant in the world, Chinese sweet wormwood, produces artemisinic acid, which is the most effective ingredient to cure all forms of malaria. Producing it in vats with bioengineered yeast would make the drug cheaper, and ultimately save lives; indeed, access to the drug could save perhaps 655,000 lives a year in the developing world. The Bill and Melinda Gates Foundation, then already campaigning worldwide to eradicate malaria, became the benefactor. The Gates Foundation granted $42.6 million to the Institute for OneWorld Health, which then partnered with Keasling’s students. Renninger still has the wire statement for the first $3.7 million. It had more zeros than he had ever seen on a check, and it was made out to a mostly nonexistent biotech company that he and his friends had called Amyris.
In March 2006, after two years of work, Keasling, Renninger, and the team had engineered the right combination of gene parts. Nothing like it had ever been done. By way of comparison, Monsanto, which has a billion-dollar research budget, has only ever commercialized a corn strain with eight new genes; Amyris had engineered 13 into yeast. And after developing the yeast, Amyris decided to give it away for free, licensing Sanofi-Aventis to sell the drug that the yeast would produce at cost. The pharmaceutical manufacturer plans to begin production by the end of 2012.
The idealism stemming from the artemisinin efforts still permeates the entire field of synthetic biology. Amyris’s success proved that Keasling’s ideas weren’t fanciful. Suddenly, people in the know were arguing that in the future, making materials and commodities—actual physical stuff—would no longer be a resource problem but a genetic programming problem. Microbes in vats could assemble all the chemicals we’d ever need. The stars of Silicon Valley venture capital heeded that call. Several firms flocked to Amyris to fund new applications of the technology. John Doerr of Kleiner Perkins Caufield & Byers, and Vinod Khosla, the clean-technology guru, made the first $20 million investment. Geoff Duyk of TPG Biotech followed. They debated making lubricants and high-value chemicals for perfumes. All of these, however, seemed to lack the nobility of battling malaria.
"Set your sights on diesel," Khosla told the Amyris team, according to Newman. "It’s the hardest thing you’d want to do, but it’s the biggest market out there, and you’ll build an incredible company." Finding an alternative to petroleum had the same ring as battling malaria: The world would be better for it.
After much debate, the Amyris team homed in on farnesene. It could be converted to a diesel-like fuel without any apparent downsides. (It did not, for instance, produce harmful particulates and sulfur when burned.) With additional chemical processing, in fact, farnesene could go into a host of other products, like cosmetics, rubber, plastics, and lubricants. These extra applications would be essential to making the company profitable within an acceptable time frame. Oil giants make their profits on minuscule margins; going after diesel would mean making a commodity that generally sells for a dollar a liter. The Amyris team hoped higher-end chemical applications for farnesene, which could sell for upwards of $20 to $50 per liter, would support the company on the long road to fuels. With farnesene, the newborn company now morphed into a strange amalgam. Amyris was a biotech, chemical, and fuel company—three different industries united by a genetically modified yeast cell. Who could lead such a company? Certainly not the founders, who’d never worked outside of academia. So in 2007, Renninger and the other founders agreed on someone else: John Melo, then the president of U.S. fuel operations at British Petroleum.
John Melo feels he was destined to be an oilman. He grew up in the Azores, a volcanic archipelago in the Atlantic, and as a 5-year-old child working on his father’s farm, he listened for a horn blast signaling that whales were passing offshore. All the men of the village raced for their canoes to harpoon whales and bring them back. "I remember the smell vividly," Melo tells me, describing the bloody husk as the men tore away the blubber—some to be bottled, some eaten. His family moved to Boston in 1973. When he finally arrived at Amyris years later, this memory of the whale hunt took on new meaning for him. The hunt was an echo of the past of the Azores, when whale oil was a vital commodity for lamps. Petroleum had made that oil obsolete. Melo wondered if biofuels would do the same to petroleum. While Renninger and the founders would jump the molecular hurdles of creating farnesene, Melo, as CEO, could handle the macro-level challenges of building a business. "What really hooked me," he says, "was the energy, enthusiasm, and the belief that you could do anything with biology—really understanding that these little organisms were refineries."
But Melo had come from a buttoned-up oil culture at odds with less-conventional ways of Berkeley grad students. The Amyris scientists balked when he tried to apply big-company rigor to employee performance. "They didn’t want to be measured," Melo says. "They said, 'Metrics? What metrics? You can’t measure this stuff. It’s science—it’s about innovation, about us having space, about us thinking. Just leave us alone, and we’ll do great science for you.'" Other differences were not just philosophical. In February 2007, on the annual company trip to Lake Tahoe, the celebration got too wild for him. "I don’t even want to tell you what they’re into," Melo recalls of the research staff. He left the raucous scene and locked himself in his room. With the sounds of the party roaring outside, he thought, What the hell have I gotten myself into? He’d abandoned a prime position in one of the world’s richest companies for bedlam, he thought. "The next day, the founders were uncomfortable; I was uncomfortable," Melo says. They never scheduled another Tahoe trip again.
Melo tried to create a road map for Amyris. Renninger had already won a number of patents for producing farnesene; piece by piece, Melo turned those accomplishments in the lab into an international company that would be centered in Brazil, where plantations produced the world’s cheapest supply of sugar, the feedstock for the yeast. By April 2010, Amyris had raised more than $156 million in private equity and had begun to demonstrate that its farnesene, when developed into fuel, met or surpassed industry standards. General Electric and Embraer, the Brazilian airplane manufacturer, tested Amyris’s jet fuel and found it no different from regular fuel. Mercedes took sample orders for diesel. They too said it was nearly identical.
By then, Amyris’s board had decided it was time for the company to go public. That way, it could raise the hundreds of millions of dollars it would need for manufacturing and distribution. The technical specs on the product now seemed excellent—a huge victory. And on paper, farnesene appeared as if it could compete, in price, with petroleum. Whether Amyris could produce the chemical cheaply at a real plant, however, was still unproven. What’s more, no one had ever built a business like Amyris’s before. Getting its plants to run efficiently, getting its yeast bug to produce optimally—it could take decades to catch up with an oil industry that had a lead of nearly a century. Also, despite the incredible biotech tools Amyris had developed, company directors knew going public meant they’d have to open the books to the market, which only cared about one question: When would Amyris turn a profit?
During the September 2010 road show that preceded Amyris’s IPO, Melo showed charts about rising fuel prices and cheap sugar in Brazil. He explained that his "no compromise" fuel could be poured straight into tanks without having to retool engines. And as a postscript, he promised that by 2011, Amyris would produce 6 million to 9 million liters of farnesene, and another 40 million to 50 million liters by 2012. That promise—those numbers—started a clock.
To hear it from Renninger, Melo’s promise is the tragic misstep of Amyris’s young and turbulent life. In his view, the company’s problems are not problems of technology but problems arising from the pitiless expectations of Wall Street. "We were chasing that number," Renninger says of the 50 million liters. Amyris would have to meet the quotas Melo had presented or lose credibility. Coming from the petroleum industry, Melo saw those volumes as laughably small. But this was a failure of comprehension. Amyris was not an oil company, and it still didn’t have a full-scale plant. "The regret is not realizing how hard it was to get the scale up," says Melo now. He soon discovered it would take a lot longer for a fermentation and manufacturing system to work than his team had estimated.
The company opened on the Nasdaq on September 28, 2010, at $16 a share. A 100-million-liter-a-year plant in Sao Martinho was still years from completion. To meet Melo’s goals, the company rented a hangar in rural Sao Paulo from an animal-feed producer called Biomin and installed two 200,000-liter stainless-steel fermenters, each the size of a two-story house. The plant, which began running in June 2011, was beset with problems. Sometimes the process worked as it had in the California labs. Other times, the enormous tanks frothed with the carcasses of exploded yeast cells.
Yet even as the plant was having technical problems, the idea of genetically engineered fuel took off in the public. Amyris became known as a biofuel company, even though it was still pursuing the chemicals market. "Biofuels capture people’s imagination. Everyone knows how big the fuel market is. But most people have no idea what a surfactant is," Duyk, the VC, says. (A surfactant is an essential ingredient in soap and cosmetics.) "People are buying a vision."
Within five months, the stock price doubled. Within a year, Amyris had announced 20 collaborations with major chemical and commodity companies around the world. "We led the IPOs, the first one out, and we were viewed as a leader in this industry," Renninger recalls. "I remember going to a conference in San Diego in January and having people come up and say, 'Hey, realize that we’re all depending on you. If you guys don’t succeed, we’re not going to have the opportunity.' "
With the deadline for hitting its 2012 quota looming, Amyris decided to create several other manufacturing facilities, including a second plant in Sao Paulo in conjunction with a small sugarcane mill called Paraiso Bioenergia. Meanwhile, Renninger’s California research staff worked to perfect the company’s technology. Their yeast still needed tweaking. Theoretically, yeast could convert 27% of the sugar it digested into farnesene. But Amyris was struggling to make a strain that yielded more than 20%.
In retrospect, Melo has second thoughts. "Given a chance to do it over again, I would focus on our Paraiso plant and getting that one up and not all the others," he says. Keasling, who founded the company with his postdocs, tells me, "Maybe [Melo] could have been safer. Maybe he could have just done one facility at a time. I don’t know a lot of the thinking that went into it, because that was kind of beyond my time. I’m not a business guy; I’m a science guy. So maybe hindsight is 20/20."
The proof, in any event, was in the price. When various Amyris plants began to yield farnesene last spring, the selling price to customers was $7.80 a liter. It was reasonable for specialty chemicals, but far too expensive for the fuel market. Only the transit authorities in Rio de Janeiro and Sao Paulo, which subsidize fuel costs for public transportation, agreed to run a fleet of buses on an Amyris-diesel blend. But there was something even more problematic going on: Amyris concedes the exorbitant price of $7.80 a liter did not cover its production costs. The faster Amyris raced to meet its own quotas, in other words, the more money it lost.
There are lessons in Amyris’s undoing. One is to see that a thrilling technology startup, no matter how innovative, is not the same as a financially successful technology startup. A month after my meeting with Renninger in California, on the first day of May, Amyris’s communications director, Joel Velasco, sent me a curt email. John Melo, the CEO, is still in charge, he wrote. But the president of global operations and the general counsel were forced out. "I should also note that Neil [Renninger] is leaving the company."
Renninger had started Amyris from a spiral notebook in his backpack. His name is listed as the inventor of its fuels. He had become CTO when he was 34. Had Velasco perhaps made a mistake? I called Renninger on his cell.
"I just heard that you were leaving Amyris. Is that true?" I asked.
"I don’t know what you’re talking about," Renninger said. "You heard what, now?"
"I heard you might be leaving the company."
"From Joel. I don’t know . . . no . . . no . . . no."
"So that’s a mistake?" I asked.
It wasn’t. On the website, Renninger’s name had been removed from the management team even before I phoned him. A few days later, Melo told me: "We need to add a new dimension that we haven’t been about, which is on the manufacturing side. It’s all about operations; it’s all about execution."
It may not be obvious what Melo means. But by firing Renninger, Amyris is choosing a path as a farnesene company, rather than a company that pursues the tantalizing (if so far unrealized) innovations of synthetic biology. Melo’s move seems likely to make Amyris a Brazilian petrochemical company, pure and simple. After all, with a narrower mission, why keep the labs in the Bay Area? The company would no longer need such an involved scientific enterprise. With a couple dozen scientists, Amyris could operate more cheaply closer to the sugarcane plants themselves.
This evolution may point to the other lessons about Amyris’s fate. Renninger cites the deadly mistake that Amyris (and Melo) made when laying out a time line for its future fuel production. And yet, that mistake seems indicative of a far larger truth—namely, how difficult it is to predict the commercial prospects of novel and complex technologies. When synthetic biologists announce they will treat microbes like tiny factories, investors and markets may be listening, but the microbes are not. Biology is not computing or engineering—at least not yet. Yeast has already been programmed by evolution. Inserting genetic instructions or working around ones already inside will be an uncertain business until scientists understand exactly how the organism functions. And that may take decades—if ever. "It’s not Google, Facebook, or Twitter," Melo says of Amyris’s business. "It’s just not that model."
In looking back on the company’s trajectory, Renninger brings up the company’s accomplishment with artemisinin, the malaria drug. "If everything around here crumbles," he says, "but we save hundreds of thousands of lives—if not a million—a year, that’s great."
But can’t Amyris be so much more? At a recent biofuel conference in Washington, D.C., Melo stood chastened before an auditorium of energy executives. "I know that about a year ago I would sit up here and tell you I could build all these things in 18 months flat, no problem," he said. But things were different now. "When you know where bottom is," he said, "you can look up and see clearly how to build."