It's easy to get ridiculously excited about solar power these days. The panels keep getting cheaper as technology improves. Large photovoltaic arrays are sprouting up around the globe. Sure, solar still produces only 1 percent of the world's electricity, but it's growing at double-digit rates each year.
So with all this momentum, you'd think the solar industry could kick back and celebrate, right? Domination is only a matter of time!
Well … not so fast. A provocative recent essay in Nature Energy by two solar analysts, Varun Sivaram and Shayle Kann, argues that solar still has some hard economic obstacles to overcome before it can become a major energy source and provide (let's say) one-third of our power. Overcoming these hurdles could mean the difference between solar leveling off as a niche technology and solar taking over the world.
Thanks to a little-discussed phenomenon known as "value deflation," the electricity generated by solar panels gets less and less valuable as more panels come online. The corollary is that over time, solar panels continuously need to get much, much cheaper if we want them to scale up significantly.
How cheap? Sivaram and Kann argue that the industry should set a goal of pushing the installed price of solar to $0.25 per watt by 2050 — down from around $3 per watt today. That's a mind-bogglingly low number, and it could require thinking about solar innovation in a radically new way. Current approaches to cutting costs won't necessarily get us there. We may need experimental new technologies. Or novel ways of integrating solar into our walls and windows. Or robot installers.
In an interview, the authors said that we're still many years away from value deflation becoming a crippling problem. But that's why it's dangerous to get complacent. If the industry doesn't start crafting long-term goals now, it could get stuck on a "incrementalist strategy [for cutting costs] that locks us into a suboptimal solution," says Sivaram, the Douglas Dillon fellow at the Council on Foreign Relations.
So let's take a closer look at their argument, which involves thinking through a variety of possible futures for solar — some quite exciting.
Why "value deflation" could pose a big problem for solar power
Sivaram and Kann's analysis starts with a very basic fact about solar panels: They generate their electricity during the day, ramping up to a peak in the afternoon, when the sun is high in the sky.
Right now this is a feature, not a bug. Electricity use also often soars on hot, sunny afternoons, so installing more solar panels can help satisfy this demand. And because solar panels have zero marginal cost — once they are installed and paid off, they generate power for free — grid operators take their electricity instead of resorting to more expensive "peaker" natural gas plants. That pushes down the wholesale price of electricity in the day. Good news for consumers!
The trouble comes when you keep adding more and more solar panels to a given grid. As that happens, the wholesale price of electricity keeps falling during sunny hours. And each additional solar panel becomes that much less economically valuable.1
Technically, value deflation is an issue for any energy source. If you add more natural gas plants to the grid, the marginal value of each subsequent plant decreases. But it's a bigger problem for solar panels because they're all generating electricity at the exact same time of day. So they eat into their own market share much more quickly.
A 2015 report from MIT, "The Future of Solar Energy," illustrated how this dynamic might play out in a competitive electricity market like Texas's.2 As solar penetration in a given grid system increases, the average market price of electricity drops slightly (red line). But the revenues generated by solar plants falls dramatically (blue line):
This is "value deflation," and it could become a big hurdle for solar as it goes mainstream, potentially limiting its reach. Modeling studies of Texas and Germany suggest that when solar penetration rises from zero to around 15 percent of a given grid system, the value of solar falls by roughly half. Studies of California have found similar effects.
Now, you could technically prop up solar in this situation by providing more subsidies or changing grid regulations to shield solar power from falling wholesale prices. But that would just create a new problem: value deflation would make these policies steadily more expensive over time The panels don't become inherently more valuable because they're subsidized. (See here for much more on this point.) One way or another, value deflation has to be dealt with.
Note that many solar projects are still insulated from competitive wholesale markets. A lot of utility-scale solar receives fixed revenues under long-term power purchase agreements (PPAs). And, under net metering rules, owners of residential rooftop solar panels are typically rebated at higher retail rates for their electricity.
But in the long term, as solar power expands, these policies are almost certain to change — many states are already rethinking net metering rules — and solar owners are likely to be compensated at something closer to wholesale rates.
Batteries can help fend off value deflation — but only partly
This isn't a surprise, and energy wonks have been pondering how to address deflation for some time now. The catch, Sivaram and Kann say, is that none of the strategies on offer fully solve the problem.
So take one possible response: If a given grid system becomes saturated with solar, it can connect with other nearby grids to ease the strain. California, for example, now has more solar power than it can handle and is trying to link with grids in Utah and Oregon. If you expand your grid, you can increase the total number of solar panels you can install before value deflation starts biting. But this is only a temporary solution — postponing the ultimate day of reckoning.
Second, utilities can deploy batteries or other forms of storage so that the electricity generated by solar panels in midday can be saved for use later, when it's more valued. You'd have to install a lot of storage as solar grows — especially to deal with seasonal swings in sunlight. Still, storage is getting cheaper, so this is a major area of interest for the electricity industry.
Third, there are all sorts of clever methods that utilities can use to shift around electricity demand, to align it better with solar output. My colleague David Roberts has written about this extensively. Smart water heaters, for instance, could take any excess solar electricity during the day to heat up their water, storing it for later.
Sivaram and Kann agree that those last two mitigation strategies are especially promising. But they may ultimately prove limited. One study of California looked at an extremely optimistic case for deploying storage, in which battery prices fall faster than anyone expects by 2030. The authors found this would alleviate just 30 percent of value deflation for solar. That same study found that smart demand management, with time-of-use pricing, would alleviate another 15 percent. Both would help … but they wouldn't eliminate the problem.3
There's a fourth option, too: utilities can retire older, less-flexible generating units that exacerbate the value deflation problem. Existing coal plants, for instance, can't easily ramp down in the afternoon when solar power ramps up. But as they're replaced with, for instance, nimbler gas plants, that will ease the daytime traffic jam of electrons caused by solar. This also only postpones the day of reckoning, but it can help.
"Storage and load controls definitely do have potential to mitigate value deflation, but they're not going to fix the problem entirely," explains Kann, the senior vice president for research at Greentech Media. "And the more germane point is that the solar industry shouldn't rely on these strategies for its future."
That last point is important: If battery costs don't fall as sharply as hoped, or if utility regulators don't make the complex reforms that can enable better demand management, then the solar industry needs a backup plan.
That backup plan is cheaper solar — way cheaper than anyone's talking about. To outrun value deflation, solar costs have to keep falling year after year. Right now, many analysts are focused on when solar will hit "grid parity" and become competitive with retail electricity prices without subsidies. "But that's just the beginning of the race," says Jesse Jenkins, an energy researcher currently studying the transition to zero-carbon electricity systems at MIT. "Solar cost reductions then have to continue to outpace their declining marginal value."
That's why Sivaram and Kann think the industry needs to set a long-term cost target of $0.25 per watt, to outrun value deflation. And, they argue, we're not currently on pace for that.
Yes, solar is getting cheaper. But it needs to pick up the pace.
"Okay, wait," you might be thinking. "Isn't solar already getting cheaper?" Yes and no.
It's true that over the past five years, the cost of crystalline silicon photovoltaic panels — the current solar technology of choice — has fallen 80 percent, as China has ramped up manufacturing and the efficiency of cells improves incrementally. In the US, the price of installed utility-scale solar has dropped from $6 per watt in 2007 to around $3 per watt in 2014.
As with many technologies, there seems to be a consistent learning curve for silicon photovoltaic panels — costs have historically fallen 18 percent for every doubling of production. So Sivaram and Kann extrapolate that forward. If the world manufactured enough panels in 2050 to provide one-third of our electricity, we could expect the installed price to drop to somewhere between $0.30 and $0.70 per watt:
That's pretty darn cheap! And it shows how far we can get simply by deploying and tinkering with existing tech. But the middle or high end of that range might not be enough to outpace value deflation.
What's more, this simple extrapolation for silicon panels could be overly optimistic. "It would be a mistake for the solar industry to put on blinders in a sprint toward silicon solar cost reduction," Sivaram and Kann write. "In decades, the industry may find it backed the wrong horse." As a safeguard, then, Sivaram and Kann argue the solar industry should explore alternative technologies beyond existing silicon-based panels — tech that could help make solar drastically cheaper down the road.
What the future of solar power might entail: perovskites, quantum dots, robots…
For example, some scientists are currently tinkering in laboratories with solar cells made from a material called perovskites. These cells have achieved stunning improvements in efficiencies in very short time frames. This technology isn't yet ready for prime time, but in theory we might one day make ultra-cheap, thin, transparent perovskite cells that could coat our shingles and windows.
That's not the only tech worth exploring. Sivaram notes that quantum dot solar cells hold the promise of being able to bust through traditional thermodynamic efficiency limits. And organic solar cells might be able to do some of the things perovskites can, only without requiring toxic lead. (See here for more emerging solar technologies.)
The catch is that all of these newfangled ideas are risky, uncertain, and quite far from commercial viability. Perovskites haven't generated much interest from the private sector to date, since the cells aren't yet durable enough to withstand real-world use and most companies prefer to focus on existing silicon panels, which are extraordinarily reliable.
Ultimately, Sivaram says, industry investment will be essential if all this advanced technology is ever to escape the lab. Academics tend to be far more interested in testing innovative solar cells under ideal conditions, rather than doing the sort of durability testing that makes sure these devices hold up in the real world.
Jenkins, who wasn't involved in the study, points out that to hit anything close to the $0.25-per-watt target we'd also likely need to rethink how we install solar panels. After all, the panel itself is only half the cost. "Balance of system" costs, like the worker who puts the thing on your roof, are the other half. That's an equally big challenge. Perhaps one day we'll have to use robots to set up arrays. Or develop "paintable" solar cells that can be easily applied to roofs, windows, and other surfaces.
How do we get advanced solar? Semiconductors offer one model.
Achieving the $0.25-per-watt goal, Sivaram and Kann argue, will require the solar industry to broaden its focus: not just ruthlessly cutting costs for existing technology but also plunking down money on future ideas.
And they don't see that happening yet. There are certainly some US companies devoting resources to innovation, such as FirstSolar, SunPower, and SolarCity. But overall, US solar manufacturers only spend about 1 to 4 percent of their revenues on R&D. (Chinese manufacturers spend even less.) Compare that with the 10 percent or higher you see in the computer and electronics industry.
To some extent, that hesitancy is understandable, explains Kann. The industry is just emerging from a brutal period — between 2010 and 2013 — when China flooded the world with cheap silicon panels and solar firms were going bankrupt left and right. "Most companies were rightly focused on battening the hatches," he says.
What's more, he says, technology incrementalism has worked for the solar industry so far. Academics have been raving about futuristic new solar technologies for years, while the industry has long insisted that existing technology was viable. "So far, industry has been right about this debate." Kann says. "Silicon technology has gotten much cheaper than most people thought it could. So it's understandable that they'd want to continue to focus on incremental improvements."
Sivaram and Kann's worry now is that complacency could set in. And they have a few ideas for boosting innovation.
Sivaram thinks the solar industry could look to the semiconductor industry. In the 1990s, 14 leading US semiconductor manufacturers formed a joint R&D consortium known as Sematech, with the aid of $100 million per year in federal subsidies. They set long-term goals and worked together on setting standards. "We're calling for something similar," Sivaram says. Solar manufacturers might collaborate with those working on innovative deployment models, and so on.
The federal government could also help here. The Department of Energy already set up the SunShot Initiative that aimed to get solar down to $1 per watt by 2020. Various experts I talked to differed on how big a role SunShot actually played in driving down solar costs, though many agreed it was quite effective at accelerating deployment by reducing permitting and interconnection hurdles.
Whatever the precise strategy, Sivaram and Kann's main point is that existing silicon technology might not be enough for solar to become a major energy player — and they should at least hedge against that possibility by pushing for other technologies as well.
Can solar power become cheap enough to dominate?
I was curious to hear reactions to Sivaram and Kann's argument, so I ran it by (among others) Justin Baca, the vice president for markets and research at the Solar Energy Industries Association, the main trade group for the solar industry.
By and large, he agreed that value deflation was a real issue, although he thought the paper downplayed some of the ways we might ultimately solve it. For instance, if countries electrify their vehicle fleets, that could create a massive new outlet for solar electricity during the day.
Importantly, Baca also pointed out that many US electricity markets remain heavily regulated, and are likely to be for many years to come. In those regulated grid systems, solar could end up being shielded somewhat from declining wholesale prices. (Though, again, value deflation would happen regardless — and would make any regulations more costly over time.)
That said, Baca did agree with the broader point that solar would have to continuously get cheaper, even if he wasn't sure 25 cents per watt was the number to fixate on. "The exact number they landed on isn't as important as the idea," he says. But Baca was also more sanguine about the current trajectory of R&D efforts: "Every manufacturer is already thinking about innovation anyway, to stay ahead of competitors," he said, pointing to FirstSolar's work on thin film and SolarCity's advanced solar plant in Buffalo.4
Interestingly, Baca also pointed out that solar innovation tends to happen in a pattern of punctuated equilibrium, with steady incremental improvements unfolding for years and then occasional big step changes. (One such "step change" was the huge boom in Chinese manufacturing a few years ago.) One takeaway there is that extrapolating from current trends can be really tough.
Jenkins framed it a slightly different way: Ultimately, the rate of cost-cutting and innovation would help decide how far solar can go. "The goalposts for solar are going to move constantly," he says. "Innovation has to keep pace. I think it can, but it has to be a concerted effort. If we can do that, then solar could become as important natural gas is" — supplying one-third or more of our electricity.
"Otherwise, there's the possibility that solar could get stuck where hydro is," he says. Hydropower provides just 7 percent of America's electricity and likely won't grow much. "Don't get me wrong," Jenkins adds, "7 percent isn't bad. That'd be a huge step up for solar! But it's also not dominant."
- I previously did an in-depth Q&A with Varun Sivaram on perovskites and the challenges of solar R&D, for those who want to go deeper into the topic.
- Jesse Jenkins has written extensively on value deflation, and why grid reforms alone likely can't overcome the problem.
- David Roberts has written some great pieces about the economics of solar (and wind) power here, here, and here. Also see his two pieces on the "duck curve" that can pose a challenge for solar.