Chapter 2 Coming of Age
In the fall of 2007, I was torn between two offers I couldn’t pass up. One was to attend Stanford University, my dream school as far back as I could remember. The other offer was to stay on as a process engineer at a Silicon Valley start-up called Nanosolar that was going to change the world. The chief executive officer, a visionary entrepreneur from Germany named Martin Roscheisen, put his arm around me and told me in his booming voice, “Varun, it’ll be just like Xerox PARC.” He was referring to the legendary Palo Alto research park where the modern computer was invented.
Nanosolar had set out to commercialize a technology that it knew was just as revolutionary. The company had invented an alternative to the silicon solar panel: its idea was to use an inkjet printer to deposit a solar photovoltaic (PV) coating right onto unrolling spools of aluminum foil. The resulting solar foil would be flexible, lightweight, and cheap. At the time, silicon solar panels—in addition to their hefty weight, rigid shape, and ugly aesthetics—were expensive, making the cost of their electricity more than twice that from the grid.1 Silicon was a costly commodity, and the process for manufacturing solar cells was adapted from that for producing expensive microchips. Roscheisen, formerly an Internet entrepreneur, planned to apply Silicon Valley’s model of disruptive innovation to upend the lumbering solar industry.
It was an exhilarating time to be in solar. Al Gore’s documentary An Inconvenient Truth had captivated the country. Investors were juiced at the prospect of red-hot market growth, running up the stock market valuations of early solar companies. At Nanosolar, the culture was infectiously optimistic. Around team lunches, we would dream about shipping cheap solar coatings to every corner of the developing world and carpeting remote deserts. We pulled all-nighters to prep prototype products for the CEO to show off to investors, reporters, and dignitaries, all of whom lapped up our vision. I reluctantly left the company to go to Stanford, but I eagerly kept tabs on the fundraising progress. In 2008, Nanosolar landed a $300 million venture capital round—the biggest of any company in Silicon Valley save Facebook.2 The sky was the limit for next-generation solar technology.
Then the sky came crashing down. The price of PV panels plunged—between 2008 and 2013, it fell 80 percent—and a flood of cheap Chinese silicon panels washed away nearly the entire crop of American start-ups. To be sure, Silicon Valley shouldered part of the blame, as companies made questionable business decisions and investors lost patience in their portfolios. In Nanosolar’s case, the firm scaled up manufacturing before it ironed out the kinks in its technology, and when the deluge of Chinese panels arrived, Nanosolar was in no position to compete. In 2013, it went belly up, selling off its assets piecemeal, for pennies on the dollar.
Amid the upheaval, though, something remarkable happened. Solar came of age. Two winners emerged from the brief and brutal price war that destroyed most solar producers across the United States and Europe. One was China, which established unchallenged dominance in the production of solar panels; it also became the world’s largest market for their deployment. The other was silicon—long considered a stand-in material waiting to be replaced by a superior one. It surprised industry insiders by remaining the preeminent solar technology and pushing aside all other rivals. This partnership between China and silicon transformed the solar industry into a global powerhouse. In 2016, investors sank $116 billion into solar projects around the world, more than for any other power source, clean or dirty.3
It’s about time. Humanity has dreamed of tapping the sun’s potential for thousands of years, and the development of photovoltaics in the twentieth century offered a way to do just that. Solar initially took a back seat to nuclear energy, the poster child of the postwar era and focus of U.S. government support. That left early solar firms to hone PV technology in remote and exotic applications, like powering offshore oil platforms in the Gulf of Mexico, phone towers in the Australian Outback, and satellites out in space. Then, as Japan and Germany took the lead from the United States in supporting their domestic solar industries, solar PV started to gain traction in the early years of the twenty-first century. China’s solar industry had modest beginnings, but when it finally got going, its competitors stood no chance.
Even though the solar industry is finally in the ascendant, it lost a vital element when Silicon Valley’s start-ups failed. The surviving firms harbor no illusion that solar technology might change fundamentally. Rather, in each segment of the industry, all the way from upstream manufacturing to downstream deployment, firms are laser-focused on cutting costs rather than disrupting the current order. This approach looks set to fuel continued growth in the coming years, but it is not at all conducive to the innovation the industry needs to pursue to brighten solar’s long-term prospects.
Three Millennia to Get Going
The abruptness of solar’s recent rise is all the more pronounced given humankind’s long history of trying, but not quite figuring out how, to turn it into a mainstream energy source. For most of that history, the focus was on the sun as an energy source for heating water rather than generating electricity. Although the silicon solar panel was invented more than sixty years ago, it is a newcomer in the context of humanity’s millennia-old quest to harness the sun’s energy.
Over 3,000 years ago, the Chinese used yangsui, or “burning mirrors,” to focus sunlight to kindle a fire—the oldest-known human use of solar energy. In the seventh century BCE, the Chinese pioneered solar architecture, designing houses that faced south to admit light and heat from the sun during the winter months, whereas roof eaves kept out the higher sun in the summer months to moderate the temperature in hot weather. Solar architecture also flourished in the Mediterranean: Socrates delivered lectures on the subject in Greece in the fifth century BCE, and in the first century BCE, the Romans invented the heliocaminus, a glass room that used the greenhouse effect to warm public baths.4
Solar energy appealed to ancient civilizations for many of the same reasons it does now. Inhabitants of the Greek island Delos used solar architecture to minimize their energy use and improve their energy security in the event that they had trouble importing needed timber. And by the fourth century BCE, Chinese city-dwellers, who had deforested most of their surroundings and thus lacked firewood, turned to solar architecture to conserve energy and in turn preserve the environment. This reasoning was not confined to antiquity. After World War I, solar architecture became popular in Germany as a way to limit fuel dependence after the Allies occupied the coal-rich Ruhr district.5
The next level of sophistication in harnessing solar energy was to convert sunlight into mechanical work. In the first century CE, Hero of Alexandria built a solar siphon that heated air, which expanded to propel water from one chamber to another. Most other inventors focused on using sunlight to heat water instead to then perform mechanical work. A particularly audacious example was a 4-mile-long mirror designed by Leonardo da Vinci to heat a factory’s boiler (he started building it but never finished).
Augustin Mouchot, a French mathematics professor, made the most impressive leaps forward in putting solar to work. He envisioned three uses of solar power that are still actively being developed today: driving a heat engine, generating electricity, and producing portable fuels. In 1874, he built the world’s first solar engine in Tours, France, using an 8-foot conical mirror to focus the sun’s energy on a boiler that drove a 1/2 horsepower engine (for reference, that’s roughly enough to run some power tools in a modern woodworking shop). Then, in 1879, he figured out how to convert solar radiation into electricity by reflecting sunlight to heat the junction of two metals soldered together, generating an electric current. He used this electricity to split water into its constituent atoms—oxygen and hydrogen—intending to store the hydrogen as fuel. None of these three applications was cost-effective, though, and Mouchot soon abandoned his research—but he succeeded in inspiring modern uses of solar energy.6
The first form of solar energy to gain commercial traction was solar water heating. Around the turn of the twentieth century, solar water heaters proliferated in southern California, offering homeowners substantial savings over using coal or gas to heat their water. California would remain an important market, and by 1977, 60 percent of the state’s pools would be heated by solar energy. In 1979, President Jimmy Carter installed a solar water heating system on the White House roof.7
International markets grew rapidly as well. By 1986, 60 percent of the population of Israel had adopted solar water heaters, as the government strongly encouraged energy conservation to limit the country’s dependence on imported fuel. Similarly, the Japanese enthusiastically adopted solar water heaters, in some cases using them to heat popular public baths. But it is the Chinese who have fueled the most recent and pronounced boom; by 2010, China accounted for 70 percent of all solar water heaters worldwide. Around the world, the technology generates over 1 percent of all energy used for heating.8
Although solar water heating has experienced a respectable rise over the last century, it has been eclipsed by the explosive growth of solar-generated electricity—that is, solar power. (Note: The term “power” is often used interchangeably with “electricity.” Its technical definition, however, is the amount of energy transferred per unit time.) There are two modern methods of transforming sunlight into electricity. First, concentrated solar power plants focus the sun’s rays to heat water or another medium such as molten salts to ultimately produce steam that drives a generator. Second, solar PV panels, typically made of silicon, directly convert sunlight into electricity. Although both methods have roots in the late nineteenth century, major advances in the twentieth century enabled their commercialization.
Following Mouchot’s invention of a solar engine, U.S. inventors made several advances that would ultimately underpin today’s concentrated solar power plants. Around the turn of the twentieth century, American entrepreneurs went to great lengths to demonstrate new collector designs that concentrated sunlight to generate heat efficiently. In 1901, a bowl-shaped reflector, hooked up to a fifteen-horsepower engine, debuted at an ostrich farm in Pasadena, California. (The contraption was billed as an oddity among oddities.) Then, in 1911, Frank Shuman raised venture capital to build rows of U-shaped solar collectors in the Egyptian desert to drive irrigation systems. Decades later, in the 1970s, an American patented a scheme in which circular arrangements of heliostat mirrors would focus sunlight at a single point atop a central power tower.
Another advance was to store some of the heat for use when the sun went down. In 1904, a solar installation in Needles, Arizona, managed to power an engine 24/7 by storing some of the hot water for later. Together, progress in converting solar energy into heat and storing that heat inspired the concentrated solar power industry. Nowadays, new solar thermal plants increasingly use the power tower configuration and have built-in heat storage so that the plant can generate electricity around the clock.9
Still, because solar PV panels have recently fallen in cost much faster than concentrated solar power plants have, adoption of the former has accelerated, while adoption of the latter has slowed to a trickle. By the end of 2016, just 5 gigawatts (GW) of concentrated solar power capacity was online around the world, compared with over 300 GW of PV capacity.10,11 The development of photovoltaics, which took place on a parallel track that also reached back into the nineteenth century, set the stage for the global solar industry as we know it today.
From Selenium to Silicon
The story of solar PV power starts in 1839, when the French physicist Edmond Becquerel discovered the photovoltaic effect. He immersed silver chloride in an acidic solution, illuminated it, and connected it to two electrodes, between which an electric voltage developed. Becquerel also noticed that his best results came from using blue or ultraviolet light, a phenomenon that he couldn’t explain.
Forty years later, an English engineer, Willoughby Smith, discovered that selenium, a material known today as a semiconductor, became more conductive when exposed to light. Researchers at King’s College London tested this discovery by exposing selenium to candlelight and then abruptly screening the candle; because the selenium’s conductivity dropped immediately, they concluded that fast-moving light, rather than slow-acting heat, was the cause of the electrical activity. This behavior was entirely mysterious to scientists at the time, but that didn’t stop an American inventor, Charles Fritts, from building the first solar panels out of selenium and installing them on a New York City roof in 1884.12
It was two more decades before Albert Einstein finally solved the mystery and explained how light was turning into electricity. In a 1905 paper that in time would win him the Nobel Prize in Physics, Einstein posited that light was composed of tiny packets—or photons—of energy.13 Sometimes, he explained, a photon had enough energy to knock an electron out of its customary orbit around the nucleus of an atom in a metal or semiconductor, and that electron could then move freely. By freeing enough electrons, a stream of photons could generate an electric current.
Einstein’s theory elegantly solved the pesky question of why only some colors of light produced an electric current. The energy of a photon depends on its color. Blue and violet photons have the highest energies in the rainbow; red photons have the lowest. Only some colors are energetic enough to kick electrons out of their orbits. Photons of the other colors sail right through a PV material. Which colors are absorbed rather than transmitted depends on the material.
But even though the theory had caught up with experimental reality, photovoltaics would remain a curious oddity for the next half-century. Writing in 1935, an executive at the American firm Westinghouse wrote presciently that selenium devices, which converted only 0.5 percent of the sun’s energy into electricity, would need to get fifty times more efficient to become practical.14 As it happens, today’s state-of-the-art silicon PV cells are 26 percent efficient, roughly fifty times better than their selenium predecessors. (A solar PV cell is simply the basic building block of a solar panel. Cells can range in size from that of your fingernail to that of your hand. The former are typically research devices, whereas the latter are used in commercial solar panels. Panels are somewhat less efficient than individual cells because wiring cells together to make a panel can lead to various power losses.)
The first silicon cells weren’t this efficient, of course. In 1953, Gerald Pearson and Calvin Fuller—researchers at Bell Laboratories who had helped invent the silicon transistor, the building block of the modern computer—realized that their device was highly sensitive to light. So they recruited Daryl Chapin, who was looking for a way to power remote telephone installations, and the trio built the first silicon solar PV cell. Although their first device was just 2.3 percent efficient, that blew selenium out of the water.15 Chapin then doggedly worked to improve the silicon solar cell. The next year, in 1954, he unveiled a 6 percent efficient PV cell to great fanfare; the New York Times reported that the advance could lead to “the harnessing of the almost limitless energy of the sun for the uses of civilization.”16
Peddler on the Roof
Unfortunately, the timing couldn’t have been worse. The advent of silicon solar technology would be overshadowed by an even more heralded clean energy technology: nuclear power. That would leave the nascent solar industry to scour remote settings on and off the planet, peddling its wares. Indeed, the industry has been on something of a roller coaster ever since.
In 1953, President Dwight D. Eisenhower delivered his “Atoms for Peace” speech to the United Nations General Assembly, laying out his vision for “universal, efficient, and economic usage” of nuclear power.17 This vision would captivate popular imagination about—and monopolize government support for—nuclear technology, eclipsing solar power. A sign of the stark contrast was the pageantry of the 1955 Atomic Conference in Geneva, which attracted heads of state. In comparison, the low-profile inaugural World Symposium on Applied Solar Energy, held three months later in Phoenix, Arizona, barely attracted media coverage.
At the time, solar PV technology was still prohibitively expensive—a 1-watt cell cost $286 to produce, implying a cost of $1.4 million for solar panels to power a single American home.18 But the research and development (R&D) funding necessary to drive the cost down was not forthcoming because all eyes were on nuclear. During the 1950s, U.S. government funding for solar R&D was limited to $100,000 per year, whereas nuclear power received over $1 billion a year.19 As a result, the use of solar PV technology for commercial power generation languished while nuclear power boomed—a reversal of modern trends.
But even as President Eisenhower pulled the rug out from under the fledgling solar industry with Atoms for Peace, just two years later, in 1955, he gave it new hope by announcing America’s intentions to launch a satellite, kicking off the space race. Solar power quickly emerged as the only solution to keep satellites operating indefinitely—onboard batteries or fuel would run out after several days of use. The Soviets were first into space with their Sputnik satellites, but when the United States launched its Vanguard I satellite in 1958, its solar-powered radio lasted far longer.
Over the next decade, as the space program ballooned, it provided a steady stream of orders for solar panels, sustaining a $5–10 million market.20 These crumbs enabled a few firms to invest in developing increasingly efficient solar panels because it was crucial to extract as much energy as possible from the limited number of panels that could be sent into space. Thanks to the space race, the fledgling solar industry slowly grew in size and sophistication.
The early applications of solar power on Earth itself were far from civilization. An advantage of solar over petroleum fuels or batteries was its low maintenance—you could leave a solar panel out for a decade or more and it could be trusted to generate power every day. That made it attractive for settings where routine maintenance was impractical. Thus, Australia used solar panels to power telecommunications repeaters and distant phone systems across the Outback; and Exxon financed its own company, the Solar Power Corporation, to deploy solar systems on its offshore oil rigs in the Gulf of Mexico in the 1970s.21,22
In fact, in an ironic twist, oil companies invested substantially in solar power at its outset. In addition to Exxon, the U.S. oil firms Mobil, Arco, and Amoco all had solar divisions. Arco made the biggest bet, investing over $200 million and building the first solar factory with an annual capacity greater than 1 megawatt (MW) in 1980; by 1988, it was the world’s largest solar producer.
Oil companies had entered the space for a variety of reasons: Solar could be useful to their business operations by powering offshore platforms or service stations; the firms’ versatile engineers were well suited to advancing the technology; and in Arco’s case, the company’s commitment to environmental stewardship drew it to solar.
But the American oil companies divested their solar portfolios by the 1990s, just as European oil producers started looking for a piece of the action. By the 1990s, British Petroleum (BP) and Royal Dutch Shell had cracked the list of the top four solar producers.23 Yet ultimately, American companies—and, eventually, European companies by the 2000s—decided that solar was too far afield of their core business interests. Many also lost money on it, and nearly every firm left the sector. Today, however, as solar market forecasts predict rapid growth, some oil producers are expressing interest once again. Indeed, in 2011 a French oil major, Total, acquired a majority stake in a notable solar manufacturer and project developer, SunPower.
The solar industry got a big boost in the 1970s from the U.S. government as America reeled from spiking oil prices. Indeed, for a decade, the United States would build a commanding global lead in solar power. Following the Arab oil embargo of 1973, the price of oil quadrupled in the United States, and it tripled again by 1979 when the Iranian Revolution spooked world oil markets.
Facing gasoline shortages and economic malaise, the Carter administration shepherded incentives for solar power through Congress. These included increased support for R&D, tax credits of up to 30 percent to install solar panels and water heaters, and regulations requiring power utilities to purchase power from independent generators like solar farms. By 1980, Congress had authorized over $1 billion per year for solar incentives, including $150 million for solar PV R&D.24 And the government’s solar push established the preeminence of the American solar PV industry, which by the 1980s accounted for 85 percent of global PV sales.25
But U.S. dominance of both the solar panel production industry and the installation market was short-lived. The combination of the incoming Reagan administration’s ideology and a crash in oil prices in the 1980s led U.S. government support for solar to plunge. In 1985, Carter’s solar tax credits lapsed, and by 1988, President Ronald Reagan and his administration had led Congress to slash R&D funding for photovoltaics by 75 percent compared with its 1980 peak.26 President Reagan was partial to nuclear power, allergic to subsidies for uncompetitive energy sources like solar, and ideologically committed to funding only basic science research, rather than the applied R&D and demonstration projects supported by his predecessor.27 Unfortunately, instead of making way for U.S. companies to privately fund applied R&D themselves, the withdrawal of government support simply drove American companies out of the market. For example, Mobil and Arco sold their solar divisions to German firms in the early 1990s.
In the vacuum created by the U.S. retreat from solar, Japan and Germany seized the opportunity to lead. In the 1990s, the Japanese government launched its own incentive programs—for example, to install 10,000 solar roofs around the country—and beefed up its R&D support. Thanks to the resulting strong domestic market and innovation, Japanese firms such as Kyocera, Sanyo, and Sharp surged to become leading global solar producers in the 1990s.28
By the 2000s, Germany had created the largest market in the world for the installation of solar power, and its firms challenged Japan’s for dominance in solar PV production. In 2000, Germany passed landmark legislation that offered substantial incentives for new solar installations—a guaranteed premium price, or feed-in tariff, for which the owner of the installation could sell solar power to a utility over the next twenty years. Thus began Germany’s Energiewende (energy transition). Over the next ten years, Germany’s generous policy support would make it the world’s largest solar market and almost singlehandedly underwrite the global PV manufacturing industry’s growth.29 By 2010, Germany accounted for nearly half of the global market for solar deployment.30 As a result, solar panel production by Siemens, Q-Cells, and other German companies boomed to meet domestic demand.
Throughout the 1990s and 2000s, U.S. firms struggled to compete, squeezed by limited government support at home to improve solar PV technology or deploy more solar power. Nevertheless, thanks to Japan and Germany, solar grew into a multibillion dollar industry. Yet it was about to have an identity crisis.
Sunburned in Silicon Valley
The global solar industry had grown on the back of a single technology: the silicon solar panel. Yet for decades, scientists and entrepreneurs had been scouring the periodic table for alternative materials to silicon that could make solar power an even more compelling proposition. This schism set up a showdown between old and new technologies.
Scientists had long considered silicon a less than ideal material for a solar cell that was in use mostly out of convenience. Recall that the Bell Labs researchers who invented the modern solar cell used silicon because of the recent discovery of the silicon transistor—it was simple to repurpose that device, which the researchers had already designed equipment to produce, to act as a solar cell instead of a transistor. But silicon was known to be a weak absorber of light—that is, a particularly thick chunk of silicon is needed to absorb the same amount of light that a thin slice of other materials can absorb. On top of this, silicon absorbs more colors of light than other materials, but this broad absorption sacrifices much of the energy contained in blue and ultraviolet photons. Hence, silicon can never be as efficient as other materials that absorb fewer colors but harvest more of the energy from high-energy photons. Finally, producing high-purity silicon, in which the atoms are arranged in perfect, crystalline order and are thus most efficient in converting sunlight to electricity, requires expensive equipment, wastes a lot of silicon, and results in brittle wafers that break easily.
Recognizing these theoretical deficiencies, researchers hunted for alternatives. One of their earliest discoveries, in 1967, was not a different material, but rather a different form of silicon known as “amorphous silicon,” which could be used to make flexible, thin photovoltaic films.31 Making the films was cheaper and wasted less of the valuable silicon feedstock, but the resulting cells were often half as efficient as traditional crystalline cells. Ironically, they also tended to get worse during the first six months of exposure to sunlight before stabilizing—an unfortunate defect for a solar material! As a result, amorphous silicon solar never took off the way its backers had hoped, although it did carve out a niche as the material of choice for flexible or portable applications such as pocket calculators.
Another solar material that found a small market niche but struggled to achieve mainstream commercial success was gallium arsenide (GaAs), which is more efficient than crystalline silicon but quite expensive to produce. In 1970, Soviet scientists made the first highly efficient GaAs solar cells, and GaAs ultimately surpassed silicon as the material of choice for space applications by the 1990s because of its high efficiency. Because GaAs comes from a family of materials that are easy to stack on top of one another to harvest different colors of light most efficiently, the highest-performance solar cells today use GaAs and its cousins in multijunction solar cells—in the lab, these can exceed 40 percent efficiency. This is the technology that, for example, has powered NASA’s rovers on Mars. Still, multijunction cells did not make much of a commercial impact on the terrestrial solar market because of their high production cost.
The most serious competition to silicon came from two thin-film materials—cadmium telluride (CdTe) and copper indium gallium (di)selenide (CIGS)—that stood out for their strong ability to absorb sunlight, the potential low cost of their manufacture, and the theoretically high efficiencies that they could achieve. Through the 1990s, these technologies waited in the wings while silicon continued to dominate the market. Researchers in the lab slowly pushed up thin-film efficiencies, but companies had little success selling panels in the marketplace. Finally, in the early 2000s, the American firm First Solar managed to rapidly ramp up production of CdTe panels, which were less efficient than silicon but also less costly to produce.
By 2006, it appeared as though thin films had arrived. When First Solar went public, it was valued at a (then) eye-popping $400 million. A Silicon Valley gold rush ensued, as venture capitalists poured money into start-ups developing thin-film solar that promised to drive down the cost of material inputs compared with that of expensive silicon solar cells. That’s when Nanosolar—the company I worked for—took off, raising hundreds of millions of dollars to inkjet print thin-film solar on rolls of aluminum foil. Even more exotically, Solyndra planned to transform flat solar panels into cylindrical tubes that could absorb light from every direction. That vision earned it a whopping $1 billion in private investment, along with a $500 million loan guarantee in 2008 from the Barack Obama administration.
Many of the investors and entrepreneurs who drove the boom in solar start-ups had previously worked in Silicon Valley’s venerated semiconductor industry. There, they had pioneered the equipment and technology to make highly complex silicon microchips for electronics. These semiconductor veterans saw the solar industry as a backwater—one that had simply borrowed decades-old equipment from semiconductor production lines to make mediocre solar panels.
One of these pioneers was my father. In 2008, high on Silicon Valley hubris, he left the semiconductor industry to start a new solar venture, Twin Creeks Technologies. Rather than choose an entirely new material, he opted to stick with silicon but try to use it more efficiently. So he and his team raised several hundred million dollars and built an awe-inspiring ion cannon—as big as a house—that could peel thin slices off silicon wafers and reuse the precious material more efficiently. The New York Times hailed the technology as “radical.”32
Then silicon suddenly stopped being precious. A glut of production capacity in China plunged the price of polysilicon, the raw material input for silicon solar cells, from over $400 per kilogram (kg) in 2008 to $50/kg in 2010.33 As a result, it was no longer necessary to economize on material costs with thin-film or thinner-silicon technologies.
Stuck with a product that cost $6.29 per watt to manufacture but only earned $3.42 per watt in the marketplace, Solyndra filed for bankruptcy in 2011, leaving taxpayers on the hook to cover a half-billion dollars of debt.34 In 2012, Republican presidential nominee Mitt Romney assailed President Obama’s bet on Solyndra as “a symbol of gross waste.”35 Over the next two years, both Twin Creeks Technologies and Nanosolar also ran out of money, as did nearly every other American solar start-up. Bruised venture capitalists slashed their cleantech portfolios, and solar became a dirty word in Silicon Valley. As for my father, he left solar behind to rejoin the semiconductor industry, where innovation was alive and well.
External shocks like falling silicon prices weren’t the only reasons that the solar start-up bubble burst. As I’ve written with my colleagues Ben Gaddy and Frank O’Sullivan, investor expectations for these companies were likely unrealistic from the beginning. The start-ups had to develop new materials, customize the machinery to make those materials, design a new solar product, and cultivate a market for that product. And they had to do it all within three to five years because of the way that venture capital funds were structured to pay back their investors.36 As a result, start-ups made costly mistakes, like racing to build expensive factories to mass-manufacture thin-film solar cells before they could reliably produce small batches of efficient cells.
Today, thin films limp along, accounting for less than 10 percent of the total market. Most of that is made by U.S.-headquartered First Solar. But even as Silicon Valley’s technology revolution fizzled, the solar industry would experience a true coming of age, courtesy of China and its prodigious production of silicon solar panels.
China and Silicon: The Dream Team
In a curious twist, China’s rise to dominance in the solar industry has its roots in the work of a scientist out in Australia. That scientist, Professor Martin Green of the University of New South Wales, is fondly regarded as the father of photovoltaics. In the 1970s, while the United States poured funding into PV research, Green salvaged equipment made to process semiconductors “from the scrap heap” in Australia to begin developing high-efficiency silicon solar cells.37 When U.S. funding for solar dried up during the Reagan administration, Green’s Australian research operation emerged as one of the best labs in the world. The scientific advances that he later made in the 1980s laid out the architecture for modern silicon solar panels. And Green’s students would take that technology with them and create the Chinese solar companies that would take over the global industry.
Still, a quarter-century ago, Green’s was a lonely voice arguing that silicon solar panels were the future. Writing in 1993, he remarked:
Since silicon introduced the photovoltaic era over 40 years ago, silicon solar cells have been the “workhorse” of the photovoltaic industry. Many, including the present author, have assumed that this has been a stop-gap measure while the photovoltaic community, with widely different levels of patience, awaited the arrival of a more ideal material.
… rather than being a stop-gap, silicon has the potential to become the ultimate photovoltaic solution … [Silicon’s] strengths lie in the presently established market position, the on-going improvements to the technology, and the remaining scope for future improvement and the sustainability of silicon-based technology.38
In contrast to the accelerated timelines of Silicon Valley start-ups that tried to perfect thin film technologies in just a few years, silicon technology matured over several decades. In the 1970s, the U.S. satellite company COMSAT figured out how to carve microscopic pyramids on the surface of silicon solar cells to ensure that incoming light rays bounced around instead of being reflected away and were ultimately absorbed. These cells were known as “black cells” because they absorbed all visible light and could thus achieve efficiencies in excess of 17 percent.39
Then Professor Green took over. Starting in the 1980s, his lab unveiled a remarkable series of improvements to silicon solar cells, pushing them to over 25 percent efficiency by the turn of the twenty-first century (see figure 2.1).40,41 Even today, top silicon manufacturers are switching their production lines over to replicate Green’s cell architecture—a design known as PERC—to boost their efficiencies.42

Efficiencies of various solar PV technologies. In the last half-century, scientists have progressively increased the efficiencies—or the percentage of the energy delivered by the sun at high noon that can be converted into electricity—of solar cells constructed and tested in the laboratory. This chart plots the increase over time in the efficiencies of silicon and GaAs, which are traditional wafer technologies; CIGS and CdTe, which are thin films; and perovskite, an emerging solar technology (this technology is discussed in detail in chapter 6).
Source: National Renewable Energy Laboratory (NREL).
In parallel, private firms also made important advances. For example, firms boosted the reliability of silicon panels, improving their economics. In 1982, Arco offered a five-year warranty for its solar panels; by the 1990s, BP was offering a twenty-year warranty.43 Quadrupling the lifetime of a solar panel meant that the panel’s power could be sold for many more years, making it easier to ultimately pay back the panel’s up-front cost. This effect was just as powerful for the competitiveness of solar PV as increasing the efficiency of converting sunlight to electricity.
There was even a Silicon Valley solar success story a decade before the thin films debacle. As industry pioneer Dick Swanson tells the story, in 2001 he was ready to commercialize the highly efficient silicon solar panels that his start-up, SunPower, had developed. But for several years, nary an investor had showed any interest in funding solar power. So the company was about to run out of money and was preparing to lay off half the company. Then Swanson reconnected with an old Stanford classmate: T. J. Rodgers, CEO of Cypress Semiconductor, a well-known chipmaker in the Valley. On hearing Swanson’s pitch, Rodgers wrote SunPower a personal check for $750,000 to prevent the layoffs. Cypress went on to acquire SunPower, supplying the investment and manufacturing expertise to scale up production. As of 2016, SunPower still manufactures the most efficient silicon solar panels on the market.
Building on this progress—including improvements to silicon cell technology in the laboratory and the accumulation of expertise and equipment in the global solar industry—Chinese companies entered the sector. Students from Professor Green’s lab fanned out across China to start or join silicon solar firms. One former doctoral student, Shi Zhengrong, founded Suntech, which became the world’s largest manufacturer of solar cells.44 And other students from Green’s lab went on to become executives at every major Chinese solar producer. In addition to importing technical knowledge, Chinese companies took advantage of foreign equipment to set up their factories. From the late twentieth century through the early 2000s, they bought turnkey solar production lines from American and Canadian semiconductor firms and later sourced equipment from Germany and South Korea.45 And through joint ventures with foreign companies, some Chinese firms honed their factories to produce reliable and long-lived solar panels.46 Not all of China’s contributions to solar were borrowed from abroad, though. Chinese producers invested heavily in making their factories more efficient than foreign facilities and wringing every last cent out of their supply chains.
In 2005, China was still a minor player in the solar industry, accounting for just 11 percent of global solar production. The Chinese government considered solar too expensive for widespread domestic use and minimally supported its deployment. But over the next five years, it enthusiastically supported the expansion of its fledgling manufacturing industry. Subsidies poured in from the central and provincial governments to depress the cost of raw materials, energy, land, and components.47 And Chinese firms enjoyed a free-flowing stream of low- to no-cost loans to scale up production and defer any need for profits.48,49
China also benefited heavily from subsidies in the developed world. As domestic solar panel production boomed, over 90 percent was exported between 2005 and 2010, mostly to European countries with generous deployment incentives. In addition to Germany’s feed-in tariff, Spain and Italy offered abundant incentives. In Spain, incentives sometimes covered half the cost of a solar installation, and by 2009, the share of solar in Spain’s power mix was the highest in the world. By 2010, Chinese companies produced nearly half the world’s solar panels, and only 6 percent were sold to its own sluggish domestic market.50
But by then, there were already warning signs that China could not continue to expand its domestic manufacturing capacity at the expense of the developed world indefinitely. Thanks in part to government subsidies for domestic solar factories, China built up massive production overcapacity all along the solar supply chain, from the raw polysilicon through the finished solar panel.51 On top of this, European countries reeling from the Great Recession yanked their solar subsidies. Germany’s target for solar deployment in 2020 sank by a factor of ten; Spain slashed its incentives by nearly half.52
Combined, overcapacity and subsidy cuts drove solar producers in China and elsewhere to wage an all-out price war to win scarce customer demand. As a result, panel prices dropped 30 percent just from 2009 to 2010; they fell by half again by 2013.53 In response, the Chinese government scrambled to create a market for the domestic industry that it had created.54 (Helping out domestic producers probably is not the only reason for China’s push to deploy solar at home; others include its desire to cut down on air pollution in its cities.) As a result, China has shifted the focus of its public policies away from subsidizing the production of solar panels to funding their deployment at home. In 2013, China dethroned Germany as the largest market for solar panels in the world, and by 2016, it accounted for nearly half of global panel sales (figure 2.2).55

Chinese share of solar PV production and deployment. Comparison of China’s rising domestic share of global PV panel production (panels are also known as “modules,” and production quantities are measured in gigawatts of power-generating capacity) with its more recently rising domestic share of global solar PV installations (also measured in gigawatts of power-generating capacity). Note that Chinese production statistics include Taiwanese production, which accounted for roughly 12 percent of global solar PV cell and panel production in 2016. Chinese firms have shifted substantial manufacturing activities off the mainland, including to Taiwan and Malaysia, to avoid U.S. and E.U. anti-dumping tariffs.
Source: Center for Study of Science, Technology and Policy; Fraunhofer Institute.
The precipitous decline in panel prices upended the global solar industry, wiping out not only the crop of Silicon Valley upstarts but also established solar producers across the developed world. Whereas in 2007, solar PV production was somewhat evenly divided among China, Japan, Germany, and the rest of the world, over 80 percent of solar production was happening in Asia a decade later, mostly in China.56
In particular, 2011 saw an industry shakeout that crippled the U.S. and European industries. Several American and European solar producers went bankrupt between 2010 and 2012, unable to compete with the flood of cheap Chinese panels. Belatedly, both the United States and the European Union sued China for violating international trade laws by dumping its panels below cost on global markets. But the punitive tariffs that they imposed on Chinese imports were too little and too late to revive failed Western solar firms.
Chinese firms suffered as well, booking negative profit margins as they tried to ride out the glut in production capacity. The extension of $47 billion in lines of credit from the China Development Bank in 2010–2011 helped keep many Chinese manufacturers afloat.57 But the largesse didn’t last, as the Chinese government pivoted to providing incentives for deploying solar rather than producing it, causing major Chinese firms to go bust or merge with competitors. For example, Suntech, once the world’s largest solar producer, went bankrupt in 2013.58 When the dust settled from the global industry upheaval, China had emerged as the world’s largest producer and consumer of solar PV, almost all of which is made of silicon. That dominance is unlikely to be challenged anytime soon.
State of the Industry
Now that the solar industry has come of age, it is poised to continue growing rapidly. In the face of relentlessly falling costs and declining public subsidies, companies are just trying to stay afloat while expanding quickly enough to keep up with global market growth. This dynamic implies that the basic structure of the industry should stay relatively stable because firms are making minimal investments in revolutionary technologies. (See box 2.1 for an overview of terms and concepts used in today’s solar industry.)
Solar Photovoltaic Power 101
At the heart of a solar PV power system is the solar panel (also referred to in the industry as a solar module). Figure 2.3 breaks out the components of a solar panel. Sandwiched between layers of glass and polymers is an array of solar cells, which are typically made of silicon and convert sunlight into electricity.

Components of a typical solar PV panel.
Source: DuPont.
Next, figure 2.4 depicts a typical residential solar system. Solar panels mounted on a roof absorb sunlight and convert it to direct current (DC) electricity. However, because the grid runs on alternating current (AC) electricity, the output from the solar panels runs through a device called an inverter that changes DC to AC. Some of that AC electricity now can power appliances in the home. And if the solar panels produce more electricity than the home needs at any given moment, the excess can be sold to the main grid.

How a residential solar PV system connects to the grid.
Source: New York State Energy Research and Development Agency.
The amount of energy per second that a solar panel can produce when the sun is directly overhead is called its power generating capacity, measured in watts (W) of electric power. A single panel can typically produce 250–350 W. And the average U.S. home solar system is roughly rated at 5,000 W, or 5 kilowatts (kW). [Note: a megawatt (MW) is 1 million watts, and a gigawatt (GW) is 1 billion watts.] Residential systems are just one use of solar power. As figure 2.5 illustrates, there are four major markets, divided by the typical size of a solar system:
- Utility-scale installations range from several to several hundred megawatts of generation capacity.
- Commercial and industrial installations typically are smaller than 2 MW.
- Residential installations are typically below 50 kW.
- Off-grid installations, which are deployed where the central power grid does not reach, can be as small as a single panel.

Solar PV upstream production and downstream markets.
Source: Images from Wikimedia Commons.
As figure 2.5 shows, the solar power sector can be divided into the downstream deployment of solar panels in these four markets and their upstream manufacturing. Manufacturing a solar panel starts with mining and refining polysilicon, which is then melted into long, cylindrical ingots and sliced into thin wafers. These wafers of high-purity silicon are then turned into solar cells, which convert sunlight into electricity. Finally, solar cells are arranged in an array and sealed together into a solar panel.
The solar industry comprises a diverse set of firms. One important category is upstream producers (or, interchangeably, manufacturers) of solar equipment, components, and panels. Another category is downstream developers that shepherd a solar project through its various stages of design, financing, construction, and operation. Some vertically integrated firms perform multiple upstream steps and even have divisions that deploy solar panels into downstream markets as well.
The cost of solar power has fallen substantially in recent years and looks set to continue doing so. The simplest way to measure cost is to divide the up-front cost of a solar installation by its rated capacity in watts. The top chart in figure 2.6 plots the cost per watt of solar installations in the United States (excluding any government subsidies). The larger the solar installation, the lower the cost per watt, thanks in part to economies of scale.

The falling cost of solar PV. Projections exclude potential 2018 U.S. tariffs.
Source: GTM Research (2017).
The bottom chart focuses on the cheapest segment—utility-scale solar—and breaks out its projected costs through 2020. By the end of the decade, the cost of a utility-scale solar installation is forecast to be less than $1/W, and the solar panel itself will account for less than half of the total cost. The rest of the cost will come from associated hardware, like inverters that connect panels to the grid, electronics, and mounting system, as well as labor costs and other soft costs, like securing permits and financing.59
Still, the cost per watt of capacity is not a useful measure to assess the cost competitiveness of solar power. Whereas fossil-fueled power plants can produce power at their rated capacity indefinitely, a solar panel can produce close to its rated wattage only under peak sunlight, around midday. Because on average there is much less sunlight over the course of each day, solar installations produce much less power on average than their rated capacity. A residential solar system in the United States typically produces only a fifth of the energy over the course of a year that it would if it continuously produced at its rated power. That figure can rise to over one-third for some utility-scale systems if their solar panels are equipped with single-axis trackers, which follow the sun as it moves throughout the day. Increasingly, developers are opting for this configuration.
A better way to assess the economics of solar power is to compare the cost of the electric energy, rather than power, that it produces with the cost of the same amount of energy from other sources. Electric energy is measured by the kilowatt-hour (kWh), a kilowatt of power output sustained for one hour [1,000 kWh amount to a megawatt-hour (MWh)]. Calculating this cost entails spreading out the up-front cost of a solar installation over the energy that it produces over its lifetime, taking into account the time value of money. The investment bank Lazard calculated that the cost of utility-scale solar was in some cases lower than $50 per MWh in 2016, comparable with the cost of electricity from the cheapest fossil fuel (natural gas).60 Still, as chapter 3 explains, even a low cost per kilowatt-hour can fall short of making solar competitive if the cost of solar exceeds the value that it provides.
Throughout all the upheaval, bankruptcies, and price swings of recent decades, the solar industry and market has grown consistently and rapidly. Over the two decades leading up to 2016, global annual PV production grew at an annual pace of roughly 40 percent. Solar now supplies more than 2 percent of global electricity demand. And despite short-term swings, long-term solar panel costs have regularly fallen as total solar production has risen—by roughly 20 percent for every doubling of cumulative production.61 This regular reduction has been dubbed “Swanson’s Law” (although Dick Swanson, founder of SunPower, will tell anyone who will listen that Paul Maycock at the U.S. Department of Energy (DOE) noticed this trend first, and Swanson just popularized it later).
From 1980 to 2001, while Professor Green was churning out new and improved solar cell designs, improvements in PV efficiency drove cost declines. But since then, performance improvements have stopped driving declining costs. Rather, those declines have happened because firms have achieved economies of scale from mass production, and they have learned to wring costs out of their manufacturing processes and supply chains.62 On top of this, the costs of deploying thousands of solar panels in the desert or a dozen of them on a roof have also fallen regularly, as installers and developers around the world have gained experience and devised clever ways to install solar more cheaply. As solar becomes more widespread, these trends should continue to drive down the installed cost of solar.
At the same time, competition will only get fiercer across the entire solar value chain. In the early 2000s, the most lucrative segments of the solar industry were upstream, especially in the production of polysilicon. But when the glut of Chinese factories producing polysilicon, cells, and panels came online, prices for all of these products plunged, squeezing profits out of the entire upstream value chain.63
Now profits for firms in the downstream deployment of solar are being squeezed as well, partly from intensifying competition and partly because of a global decline in incentive payments for solar power. Historically, governments have supported the deployment of solar in three major ways. One, they have used mandates to require utilities to procure a certain share of their power from renewable sources such as solar power. Two, they have subsidized the construction of solar installations, for example through tax credits that reduce solar’s up-front cost. And three, they have paid for the energy that solar produces at a premium rate. For example, through feed-in tariffs or another policy known as “net metering,” homes and businesses sell power to the grid, often at higher rates than the power would fetch on the open market.
Feed-in tariffs in particular have been central to spurring the deployment of solar power, and in 2015, they were the most popular policy instrument to promote renewable energy in the world.64 But around the world, countries are replacing feed-in tariffs, which guarantee the same, high payments to every supplier of solar electricity, with reverse auctions, in which developers bid against one another to offer the lowest price for which they will agree to sell solar power for the next fifteen to twenty years. By pitting firms against one another, this policy shift has dramatically reduced the contract price for which developers can sell their power, squeezing their profits.65 Even Germany, which pioneered feed-in tariffs, has switched to reverse auctions, which is fiscally prudent for Germany, but disastrous for the profit margins of solar developers.
Even as profits are being squeezed across the solar value chain, global prospects for solar deployment around the world have never been stronger. A string of record-low prices for solar projects around the world has encouraged governments to set ever-more-ambitious deployment targets. Beginning in Dubai in 2015, with a then-unheard-of bid by a Saudi developer to build a project and sell power for 6 cents per kilowatt-hour (kWh), subsequent government tenders in Peru, Mexico, and Chile plunged prices below 3 cents/kWh. Then, in late 2016, Abu Dhabi seized the title of cheapest solar at 2.4 cents/kWh (figure 2.7).66 Then, in late 2017, the kingdom of Saudi Arabia seized the crown for cheapest solar at 1.79 cents/kWh (possibly with the aid of incentives).

Recent prices for solar PV power purchase agreements (PPAs) around the world. Each circle represents the price for which a developer of a utility-scale solar project has agreed to sell electricity through a PPA contract with a customer (for example, a utility that will buy the electricity and distribute it to homes and businesses). The size of each circle represents the power-generating capacity, in megawatts or gigawatts, of that particular solar installation. The shading of each circle denotes the region of the world where the project has been, or will be, built. The dashed line approximates the average downward trend of PPA prices over time.
Source: GTM Research (2017).
Recognizing ongoing cost declines, market analysts are bullish about solar’s deployment prospects on every continent save Antarctica. Bloomberg New Energy Finance projects that by 2040, the cost of solar PV will plummet by two-thirds, and as a result, solar will account for 17 percent of total electricity generation (see figure 2.8). This would be especially impressive, given that electricity demand is projected to grow rapidly in emerging economies, driven by urbanization, economic growth, and the rise of new sources of demand such as electric vehicles. So, if Bloomberg’s forecast is right, solar will grow by more than 1,500 percent through 2040, and more than 40 percent of that growth will occur in China and India, which are poised for explosions in solar PV deployment. Strikingly, these installations will come in all sizes, not just in the form of massive solar farms. The 4,500 GW of solar in 2040 is projected to break down 70–30 between utility-scale solar farms and distributed solar installations that span off-grid projects and residential and commercial rooftops.67

Projections for global solar deployment. This chart plots the expected growth in the installed solar PV capacity (measured in gigawatts) in major countries and regions around the world between 2016 and 2040.
Source: Bloomberg New Energy Finance (2017).
Still, the scarcity of profits and recent upheaval in the industry has instilled in firms a conservative, low-risk approach to the future. Indeed, the solar industry today is a far cry from the vision that Martin Roscheisen sold me on when he put his arm around me at Nanosolar and spoke glowingly of the coming technology revolution. Today, barely any investment is flowing toward post-silicon solar materials, as the existing industry structure, organized around silicon, crystallizes. Polysilicon producers are focused on maximizing the purity of silicon. Ingot and wafer manufacturers then ensure that no scrap is left unused as they deliver increasingly thin silicon wafers down the chain. And cell and panel manufacturers are intent on nudging up the efficiency of their products while minimizing the expense of the materials and factory operations that go into making them.
Downstream, developers and financiers of solar installations fixated on meeting falling cost targets will trust only existing silicon technology that has been thoroughly vetted. Solar industry associations trumpet their road map for a steady upward march of solar panel efficiency as a sign of the industry’s innovative zeal.68 But the reality is that the average efficiency of a silicon solar panel will increase each year for the next decade simply as more producers switch from low- to high-purity silicon and to more efficient cell architectures that have already been invented.
To be sure, the industry will continue to see incremental innovation. Inverters, which connect solar panels to the grid, will get smarter and less expensive. Project developers will devise ingenious configurations for how to lay out and mount solar panels to minimize wiring costs and electrical losses. Companies are already using drones to aerially monitor the operations of solar farms and cut down on labor costs.
Yet the fact remains that the industry spends on average just 1 percent of its annual revenues on R&D—a measly sum in comparison with most advanced industries. In 2017, researchers at Stanford University reported that Chinese solar companies are beginning to spend more on R&D collaborations with Western researchers and coinvesting with the Chinese government in new technologies. But it is still the case that most of these firms’ R&D expenses remain targeted at incrementally improving existing silicon technology. Technology disruption is simply not the aspiration for most Chinese firms. Rather, their preferred route to a better market position has been vertical or horizontal consolidation.69
It’s clear, now that China and silicon have won the day, that the solar industry plans to steer clear of disruptive innovation. The industry has come of age. Yet, that dearth of innovation should be deeply worrying to anyone hoping for a solar-powered future.