It all made sense. On paper.

OptiSolar, the mushrooming solar outfit that seemed destined for success just weeks before its dotcom-like implosion, had raised millions – at least $322 million according to a review of Securities and Exchange Commission documents – to realize a vision of solar manufacturing laid out nearly five years ago in a paper published by the National Renewable Energy Laboratory.

The paper by Marvin Keshner, a Hewlett-Packard scientist who became OptiSolar's CTO, and Rajiv Arya stated that the cost of solar could be radically dropped by building "Solar City" factory complexes capable of churning out 2.1 gigawatts to 3.6 gigawatts of solar cells a year. These factories would cost $500 million to $600 million and be composed of factories-within-factories focused on different tasks: an onsite glass making outfit capable of cranking out 30 million square meters of glass a year; a solar cell unit with 100 identical manufacturing lines; and a full-fledged packaging facility.

An attempt would be made to recycle chemicals – particularly silane – in the manufacturing cycle as much as possible.

Manufacturing know-how would be imported from the television and computer industries, and even from household construction. Rather than go with traditional frames, the paper suggested utilizing aluminum frames from regular windows and backing the panels with inexpensive silicon or plastic.

Modules would cost 0.60 cents to 0.52 cents per watt, nearly half the best prices available today. Fully installed, solar power would cost $1.00 to 0.88 cents a watt.

"This breakthrough in the price of solar energy comes without the need for any significant new innovation.  It comes entirely from the design of a very large, dedicated and optimized factory, the design of manufacturing equipment for a very large factory and the cost savings resulting from operating at such a large manufacturing scale," the paper stated. "Like the DRAM industry of the middle 1980s, we will see large companies invest around $0.5 B in capital to create large, optimized and automated factories that will achieve dramatically lower costs for solar panels."

Because Keshner worked at Hewlett-Packard at the time, the computer giant had first crack at it, according to sources. Then HP passed on it so OptiSolar licensed the existing IP from HP and became a start-up in 2005.

The ideas from the paper largely formed the company's business plan. OptiSolar sought patents for industrial showerheads to reduce the silane wasted in conventional manufacturing processes from around 85 percent to under 20 percent. Other patents described how panels could be supported with posts and simple mounting guides.

After building a factory in Hayward capable of producing 30 megawatts to 50 megawatts, it landed $20 million in tax breaks in 2007 to build a factory at McClellan Air Force Base in Sacramento County. By 2011, the million square foot facility would employ 1,000 and put out over 600 megawatts worth of solar panels a year, the company said then. Although the original paper discussed ways of making cheap solar panels out of CIGS, cadmium telluride or amorphous silicon, OptiSolar focused on silicon because, among other reasons, of its far wider availability.

Plans were also being laid to build an even larger factory after McClellan that would contain the in-house sub-factory for glass making as discussed in 2004.

But OptiSolar didn't stop there. Once organized as a company, OptiSolar also incorporated other ideas for cutting costs. Instead of concentrating on manufacturing solar panels, the company planned on installing them itself (through a subsidiary called Topaz Solar) and selling the power to a utility. By acting as a vertically integrated company performing several functions, the idea was that the cost could be reduced because profit margins wouldn't be split among several companies, said various sources from inside as well as close to the company. (OptiSolar revealed few details about itself in its heyday and the decline has made sources even more sensitive. Thus, nearly everyone who spoke insisted on anonymity.)

Although barely started, success followed quickly. In 2007, it won contracts to install over 200 megawatts in Ontario. The crowning achievement came in 2008 when the company won a deal to build a 550-megawatt solar farm near San Luis Obispo for PG&E over several other bidders. Toward the end of 2008, California Governor Arnold Schwarzenegger showed up for a factory opening that 60 Minutes covered.

And OptiSolar also avoided VCs, realizing early than some others that the high-capital requirements would exceed the capabilities and sensibilities of Sand Hill Road. Instead, the funds came from private equity firms accustomed to the long slog of energy deals. Many investors like Robert Puchniak and chairman Geoff Cummings came out of Canada's oil industry.

Money, tax breaks, contracts, connections, seasoned investors and a manufacturing strategy that would borrow from the hyper efficient computer industry.

So what went wrong?

Those closest to the company chalk it up to the credit crisis. When the crisis hit, the company was outfitting the McClellan factory and seeking another $200 million in funding.  Prior to the credit crisis, the plan was to hold an IPO in 2010.

"If it wasn’t for that, they might have been big," said one person.

Others, though, said that the expansive goals seem to be giving the company mission creep.  It was negotiating deals for tracts of land while also planning out factories and buying expensive equipment. Besides the already announced deals, it had secured rights to build power plants on 136,000 acres, enough for 19 gigawatts of power.

Some have compared the company’s situation to what happened to Ausra, Miasolé and eSolar. Those three companies each raised tens of millions of dollars, but found themselves in a financial squeeze after spending the lion’s share on outfitting their factories.

Employment was exploding too. The company had around 13 employees in June 2006 and 70 by the end of that year. By the end of 2007, it employed 200. A year later, it had 600 employees. One interviewee stated that the company, nonetheless, remained guarded about its plans.

And when would the revenue come in? OptiSolar, after all, was not concentrating on selling solar panels to third parties. It planned on selling solar panels to itself for its own solar power plants. Power delivery on the PG&E plant wouldn’t even begin until 2011.

The comparative efficiency of the company’s solar cells was also an issue. The company’s initial solar cells, out of Hayward, were 5 percent to 5.5 percent efficient, depending on whom you ask. That was close to the company’s target. The current industry average, however, is around 6.5 percent, according to Shyam Mehta, senior analyst at GTM Research, and thin-film cells from cadmium telluride are even more efficient. OptiSolar was planning on boosting that to 7 percent to 8 percent out of McClellan.  

One can also argue that it was unrealistic to compress its goals into a few years. Although Applied Materials’ Charlie Gay and others have talked about how manufacturers will one day own multi-gigawatt-scale facilities, no one has anything even close to that right now. Suntech Power Holdings is the only company with 1 gigawatt of capacity (see Suntech Boasts 1GW Capacity Amid Tough Times for Solar Market). First Solar and Q-Cells will only hit that milestone this year.

Without those big factories, amorphous silicon is still somewhat far away from the $1 per watt installed outlined in the paper. Single junction module cost around $1.80 to $2 a watt with a full installed cost running at $3.75 to $4.25, according to Mehta. First Solar announced in February that it has hit $1 per watt for a module, but that doesn’t include installation.

What was PG&E doing with a startup anyway? Why wasn’t it working with an established company? An independent energy consultant, Merrimack Energy Group, said that both Merrimack and PG&E were "somewhat concerned about the risk that the sponsor will not be able to produce the required components in a time frame necessary to meet the proposed 550 megawatt capacity levels and in-service dates," phase-in provisions mitigated the risk.

Other sources added that the nature of the contract mitigated PG&E’s risk. If OptiSolar failed, a substitute could be found.

The PG&E contract, though, did seem to goose investment. By the end of 2007, OptiSolar had culled $92 million in investment, according to SEC documents.  On April 15, OptiSolar filed a Reg D form stating that 38 investors put in $132 million more.  On April 24, word on the PG&E deal leaked out. $98 million more followed in the summer.

If the rise was fast, the fall was even more rapid. A few weeks after the factory opening by Schwarzenegger and the appearance on 60 minutes, the company cut 300, or half, of its employees (see OptiSolar Lays Off 300, Half the Staff).

"We simply couldn’t sustain the level of aggressive growth," said spokesman Alan Bernheimer after the first round of layoffs. 200 more got the axe in March.

Equipment originally destined for the McClellan factory remains in shipping crates.

The company’s legacy, ultimately, be the further expansion of First Solar (see First Solar Buys OptiSolar’s Power Projects). The relentlessly efficient solar maker bought OptiSolar’s utility and land deals for approximately $400 million earlier this year. The PG&E, Canadian and other unannounced deals, however, will be completed with First Solar’s cadmium telluride solar panels, not amorphous silicon.

Although some early OptiSolar backers are trying to find new investors, the equipment from Hayward and Sacramento is also being offered up for sale. 

Join industry leaders and influencers at Surviving the Shakeout: Greentech Media’s 2009 Solar Industry Summit in Phoenix, Ariz., April 14–15.

Tags: ausra, california, factories, hp, miasole, NREL, nrel, optisolar, solarcells, thinfilm