 Good afternoon, everyone. Thank you, John. I'm really delighted that what I said about 25 years ago stuck as well as it did. And it's a pleasure to be here. I look forward to sharing some of the history of solar with you, but more importantly, talking a lot about where solar is going. The main message that I have today is really about the speed of change. So early on, this is Easter, 1900, New York City. This is a photograph of folks going up and down Broadway. These are horse-drawn buggies. And there's one car on the street there. This is 1900, 1913. All cars except for one horse and buggy, which is still left. It's a period of time where change happened, a way people lived changed rapidly. I grew up on a farm. We had workhorses on our farm. I think my father was the last person in California to shift to get a tractor, which was a major cultural change for us. And the speed of change that's happening in solar is affecting the way we live our lives. It's affecting how fast we can respond to the cost reduction that has happened as solar has scaled. The benefit of solar is that we've always had a market. We've had markets early on, as John mentioned, in providing power to satellites, then line of sight, microwave repeaters on tops of mountains, and rural villages, which was what John was describing in India. And now I don't have to travel anywhere near as much internationally as I have for most of my life. So that freedom and what solar represents is the ability to give customers a choice. It's a chance to be independent of the grid or to do as we wish. A little bit of history here. The Smithsonian has an exhibit. If you get to what's now my neck of the woods, Washington, DC, the Smithsonian has a exhibit on the history of solar, actually, for a couple of years. I've been at DOE now for eight months before joining. I got together with all my old cronies and collected a lot of samples of things that they had in their garage. And now that's part of an exhibit at the Smithsonian that'll be there until November of this year. And this is a picture of the very first solar cell. It's a little piece of silicon, about a millimeter in diameter. Again, on one end, P on the other. And Russell Oll, who demonstrated the foldable take effect. There's a little curved reflector that's since corroded there. But he got a signal out of that, and that's the first silicon solar cell. And it's been silicon that has scaled the majority of the industry. What moved things along between 1940 and 1954. So this is, say, January of 1954. Most of the research in the 50s was done at corporate labs, at places like RCA and Bell Labs. General Sarnoff, who was chairman at RCA, was looking for a way to make use of the beta rays coming off of radioactive waste. And discovered that you could take a silicon device, a PN junction, collect those electrons, and use that to generate electricity. So Sarnoff talked about how RCA was gonna change the world, making use of all of that radioactive material in some constructive fashion. Bell Labs had been doing the majority of the work with silicon. And in April, following this, they announced that they had developed a more efficient silicon solar cell, which was this one centimeter diameter piece of silicon, an n-type wafer with a boron diffusion on it, back contact. Essentially a small version of what SunPower mass produces today, where the advanced lines are 24% efficient cells coming off of a mass production facility. Sizes increased, performances increased. There's a lot of learning that has happened as a result of having markets that could use the products being made as the cost came down. It allowed the scaling of manufacturing. Scale has helped drive most of the cost reduction in solar. By the way, if you're a scientist, it's very unlikely that science gets reported on the front page of the New York Times. But then again, this is sort of a new era that we're living in. The guys who invented it, Chape and Pearson Fuller, did this with little pieces of silicon that were actually cut lengthwise out of the first single crystal materials in order to get enough size. This was interesting, it was used a little bit for repeater systems. Several of the folks, Vernevron Braun, came over from Germany after World War II to develop missiles. And Hans Siegler came over from Germany after World War II to head the Army Signal Corps. It's the Army Signal Corps that developed most of the uses for silicon solar cells very early on in their scaling. Until in October of 1957, the Russians launched the Sputnik satellite and everything changed. And we needed to up that performance. The way that was done was to put solar cells on a satellite called Vanguard. After three launch attempts, Vanguard finally was lifted off. And that one tenth of, 100th of a percent watt is what powered the satellite for about seven years. Then the time that I got into solar was at the time of the first oil embargo. My peer group, the A students, all became nuclear engineers. Being a slacker as I was, I decided I'd try something different and went into solar. This is a plot that one of my first bosses at Spectrolab made, because by 1973, 1974, there had been enough satellites launched that you could get an estimate of how much the cost could come down, depending on how much volume you made, what experience there was. This plot, I actually have the original of that plot, is what the industry has followed ever since. For every cumulative doubling of the performance, the output of cells, the costs have come down 20, 21%. In a predictable way, a manufactured good, a product like a solar array, a wind turbine, those are manufactured energy sources. When you manufacture, you get experience with experience can come the opportunity for cost reduction if there are markets. Unfortunately in solar, we've had those markets available to us as time has gone by. The first modules that after leaving the aerospace industry was hired by Atlantic Ridgefield to set up R&D and manufacturing operation in Southern California. Arcosolar scaled by 1980, we were the largest manufacturer of solar panels. I'm pleased to say that there are solar panels just like this one built in 1980 out at Natural Bridges National Monument that have the same output today as they did when they were produced. There's an existence proof. If you design this well, so UV light, oxygen heat, thermal cycling day and night, don't degrade the interconnects. The fatigue life of the interconnect was well designed. We have something that can last and can last a long time. So this ultimate capital good can realize the value that it can bring. The first large scale power plant that we built for us large scale. This was a one megawatt system in the high desert, Mojave Desert just north of San Bernardino, an area near Hesperia. Southern California Edison has a substation called Lugo. It's barely visible in the corner of that. This is when the notions of being able to make better use of existing wires by putting generation out in the edges of where the wires are. And a lot of the population overflow from Los Angeles moved north of the high desert to Lancaster, Palmdale, Hesperia. And so as the need for electric power increased, the ability to put power close to where you needed it deferred into the future, the capital upgrades of adding bigger wires or more wire. In a typical grid today, roughly speaking, 40% of the capital is in generation. The rest of it largely is in wires and writes away. So we can make better use of the existing capital that we have by putting the generation out on the edges of the grid, sort of like uberizing electrons here. We have information available to us. We can send them and use them where they're needed and make better use of the existing infrastructure that's already been invested in. I love this, that Bill Gould, who is the chairman of Southern California Edison came to the dedication ceremony. We were real pleased, it took us one year to build this one megawatt. Today, PV power plants, you do one megawatt before breakfast break. It doesn't take nearly as long because we've gotten a routine, the factory in the field, the automation has come along with it. Bill said, my predecessors in this role spent their careers putting in wires. My successors in this role will be able to make use of those wires by using solar out at the edges of the grid. His son, Bill Gould Jr. runs a company called Solar Reserve. His son, Wayne, runs a venture capital fund investing in renewables. So Bill was not only prescient when it came time to look at how the grid could improve benefit from having access to solar, but he helped assure the legacy of that as his kids went into the space. So we talk about in the early 2010, 2011, expectations are that the need for electricity will continue to grow every year. It's led to folks thinking about how many more gigawatts of power are we gonna need to have to put on the grid? And as the scale of solar has expanded, the ability to get the costs down has followed and the ability to better utilize the source of generation is available here. I wanted to now just take one step back and talk about the things that are changing and how fast they are changing. One of the first things that has come along in the mid to late 1990s are variable frequency motor drives. Historically, the loads on the grid have been linear, like toasters. There's a resistor. Soon as variable speed motor drives went on the grid, now there are non-linear responses of loads creating the need for the grid operator to understand how best to use the grid. How best to use the grid. The grid operators have been working with variable input to the grid caused by non-linear loads for several decades. A lot of folks challenge solar today. Our costs have come down, wind as well. You guys are putting variable generation on the grid. You're creating havoc with the performance of the grid. Grid operators have been living with havoc a long time and increasingly so. There are very few non-linear loads, almost every device today has got some version of some form of power electronics on it. There's a feedback loop that's going on to the grid followed by compact fluorescence. Now there's a variety of things changing and I wanna point out, of course I've made the chart so you could allege that it's confirmation bias but a lot of what's changing is how many things are changing now and how fast those changes are coming. So smart meters, power management units, automated meters on the grid, having wind and solar come onto the grid, creating the opportunity for using renewables in a great way, having the internet be able to provide us a lot of information. Heat pumps probably deserve to be a little earlier on this chart, role of power electronics and we're already seeing storage, battery storage for electric vehicles in the range of $235 a kilowatt hour. That's a number that most folks said wasn't gonna happen until 2020. So a lot of the rate of change and it's as humans relating to exponential change is tough to internalize. So the point I wanted to make from the horse and buggy days is we're living in a world which is changing ever more rapidly and a lot of those changes we may take for granted because we don't think about them as having the rate of change as they do. The learning curves for solar, learning curves for building integrated circuits, the line here is a 28% slope for every cumulative doubling of chips that are made, the costs have come down by 28%. This also holds true for broiler chickens. More people have learned how to mass produce something, those costs keep coming down. So when we make things, and it doesn't matter if it's a high tech semiconductor material or it's a broiler chicken, that learning curve is very relevant to enabling a lot of those business opportunities that were on that prior chart. At DOE, our solar program, there are basically two broad themes. One full of voltaics, the other one concentrating solar, larger rays in the field that focus light with mirrors onto a hot box, a receiver, that then heats up a working fluid, usually molten salt. And those systems typically cover large swaths of the desert and they're very much like traditional power plants and that thermal source is used to run a turbine that is used to power the grid. And in the case of concentrating solar, you get to have storage along with the generation. In the case of full of voltaics, we need to be looking for storage opportunities or better ways to utilize the grid itself. In the concentrating solar, I put this in here because there's a tremendous change happening in the design of turbines. Typically today, minimum sized turbines that are cost effective are on the scale of about 300 megawatts or so. And the work that's been funded by DOE and the concentrating solar program, the fossil energy program, the nuclear program, we're all interested in using super critical CO2 with a 3D printed turbine assembly that can be about this size and be 10 megawatts. That means two important things. One, we can actually get to higher efficiencies with super critical CO2, but we can minimize the capital bite size chunk to be more palatable because we've got smaller building blocks that we can assemble and work together. I showed this because my entire graduate student period of building equipment 10 years, measuring the properties of CO2 at the critical point and I never thought it'd be relevant to anything. And now I'm in a place where actually what I learned there was useful. So many of you sitting in class may wonder what I'm studying is useful. Don't give up, hang in there. Sooner or later, it'll come around. In the case of full of voltaics, the history of evolution has been very rapid, much of the progress that's been made is sort of highlighted with a separate set of dots. But DOE has supported the R&D and more recently here, perovskites that have jumped from 10% to 20% in sort of leaping a tall building in a single bound have changed a lot of the thinking processes that people have around the role of new materials. It's important to recognize that time-tested performance is really critical to the financing. The largest single cost in deploying solar is the cost of money. And the more confidence that we can have in the performance out in the field, the lower the risk, the greater the predictability, the lower the cost of money for being able to finance those power plants. So even though new materials come along, it's gonna take some time to be able to prove that what we do in the way of accelerated tests is actually relevant to what Mother Nature does in real life. In the US, we now last year deployed a little more than 14 gigawatts of solar. And at utility scale systems, the total cost of an installed system is on the order of a dollar per watt with the module contributing about 40 cents and the rest of the system contributing about 60 cents. That rate of progress is enabling solar to scale because the generation cost is now in the ballpark in a place like Kansas City, not a whole lot of sun, to be around seven cents a kilowatt hour. In the US, about 1% of our electricity comes from solar. Here in California, about 12, 13, maybe 14% of our electricity comes from solar. Some of that from concentrating solar, the majority of it coming from full of old tax. An important dimension of what's happening here is just how many jobs there are across the entire value chain of solar. We add more than 1,000 new jobs a week. 1,000 brand new jobs a week. Those jobs are primarily here in the developer and installer portion of the industry. The average wage across the entire US for that portion of the jobs is $26 an hour. So we're in a great spot with job creation. Those aren't exportable jobs because our rooftops are here in our own backyard or our open fields are here that we make use of. The projection for new job creation in coming years, ratchet up again over the last four to five years that job growth rate has been faster than the creation of many of the industry's jobs that are going into this field. And we have these jobs all across the US, all the way north to Alaska, where a lot of folks use solar with microgrid systems in the remote areas hybridizing PV with diesel gen sets. The knowledge that we gain about how to put solar into the grid here in Palo Alto with Palo Alto Municipal Utility is just as relevant as putting solar out in that microgrid because these are the kinds of things that can be operated with a smartphone. You can have an app that makes it easy to manage integrating those sources, along with the rapidly declining cost of storage creates opportunities, not just in our own backyard. But for me, the reason I went into solar largely was to help people who didn't have power. And the same kind of microgrid concept that works here can work in Lusutu or parts of India anywhere. And so the opportunity to see solar scale relevant to both urbanized developed parts of the world as well as developing parts of the world is an important driver. During the course of my life, the population of this planet has almost tripled. That's something to me. That the scale of the number of people who need power to be able to live lives easily, to be able to get access to clean water and have some electricity. Those are fundamental things that we can address, at least in the case of electricity, with what we can do with solar. In more recent times, this is the learning curve for solar. There have been ups and downs. Most of the ups and downs up until 2005 related to the ebb and flow and the availability of silicon. Typically, each year, 30,000 tons of hyper pure silicon, 11 nines pure, are used by the integrated circuit industry to produce integrated circuits. That 30,000 tons is the same number as it was a decade ago. But the amount of silicon needed for solar today is on the order of 350,000 tons. So balancing the output of a polysilicon plant is very much like a refinery. It takes a lot of infrastructure, sources of steam, hydrogen, be able to make hyper pure silicon. That's the main driver of the industry. And today you can find modules for sale that are in the ballpark of 35 cents a watt and folks talking about being able to get below 30 cents a watt in the coming year. Much of that is a result of the fact that the total market for solar this year will probably be something on the order of 85 gigawatts. The total manufacturing capacity is about 110 gigawatts. When the auto industry got started, there were over 2,000 producers of cars. There's a great story of Warren Buffett carries with him a little piece of paper that has the list of all 2,000. Folks ask him, what should they invest in? And his point is, he keeps that both to remain humble, but to realize it's very difficult to predict the future, as Yogi Berra said. And what we're looking at and what Warren Buffett used as a rubric was not to try and pick the winner, but to identify the losers first and know who to bet against. So we're at a stage where the manufacturing capacity overwhelming the market means that there's still gonna be a lot of competition of ideas, of sources of financing, being able to leverage one way or another to optimize a particular business in the environment of tremendous competition. These falling module prices have helped scale solar and the job creation in the US because they're downstream jobs that can grow faster with components that are less expensive. In the SunShot, the Arun and Steve Chu, the team at DOE in 2011 created the SunShot program. The goal of this was to get to six and a kilowatt hour by 2020 and we're well on our way. So when I got to DOE in late August, said maybe we ought to lower that limbo bar from six cents to three cents. So we looked at ways of being able to work on improving the performance of modules, lowering the balance of systems hardware costs, increasing the life of a module, lowering the cost to install a system so that we could take where we are today at seven cents in Kansas City, that's without the investment tax credit, and take it all the way down to three cents a kilowatt hour in Kansas City by 2030. And at that level, we're able to be competitive with wholesale electricity prices in the wholesale market. There are a lot of alternative pathways, all of which get to that levelized cost of electricity. Sort of a reference point, if you say this is gonna last for X number of years and I have to spend money today at a certain interest rate, I can calculate how many kilowatt hours I'll get. So I spent X dollars and I got that many kilowatt hours levelized cost of electricity. It's a simple rule of thumb. That now needs to be much more elegantly fleshed out so that the time value and the location value of electricity are also quantifiable and maybe available in the marketplace to allow for the competition in the market to take advantage of a variety of pathways for lowering of cost. So that by going from six cents to three cents, we could increase the ability of solar to be competitive in the grid and if storage costs come along, then we can even increase that further. But this all assumes that the whole world is about solar and there's a lot of competition for the supply of electricity. So working on the time value and location value gives us an opportunity to think about where are the places to get a good foothold in order to be able to scale the industry. There's a lot of competition for pricing. Pricing is a function of scale, but roughly speaking, this $60 per megawatt hour, this is six cents a kilowatt hour. They give you kind of a sense of where price quotes are today. So even though the average in Kansas City is about seven, you can find people depending on where you are and what the cost of money is who can provide price quotes that have a very wide range. And we've already reached the place here in California where on March 11th this year, the value of solar went negative. So in California, we think about summer, we wanna run the air conditioner, output of solar panels is very high so you can help bring down that curve. And we've brought it down so far so fast that curtailment of the use of PV is the way to not lose money having PV on the grid. So better understanding what we can do in that grid and essentially this region of over generation that the California independent system operator mapped out in I think it was maybe 2013 or so actually and said that by 2019 or 2020, we'd have to worry about that over generation. And part of the worry is that the sun goes down but people come home in the evening, they turn on televisions, lights, business is still running. So it's this ramp rate towards the end of the day that receives a lot of attention for how are we gonna be able to help offset and help levelize what that looks like. This curve is referred to by many as the duck curve. That's where the duck curve metaphor came from. I learned a lot about PowerPoint graphics while making these slides. So the grid now has a lot of options for generation and transmission through substations. And now electrons can go two ways. If we have information available to us to be able to utilize solar, other distributed resources and assets in smaller scales and networking them more efficiently to make better use of those wires. And as I said, be able to control them or have an app that automatically responds to pricing signals as people become more sophisticated about time of day and location value. And one of the reasons that I joined DOE is that, well, I've had a great career in solar because what I learned in the first two years of working in this industry, I didn't have to learn anything new for the next 35 years. I was just trying to be able to remember the mistakes that I had made so I could make sure that I was able to do that. Like new ones instead of repeating the old ones. Now being able to integrate solar into the grid is essential for the scaling of the industry to have that market. And our partners in this are buildings, energy efficiency, and our partners in this are in the office of electricity in DOE. When I introduced myself at DOE the first day, folks said stand up and explain who you are. I said, I'm shortly gay. I've been in solar longer than the DOE has existed. DOE came about in 1977 or so. And those departments have ossified as time has gone by. So it's important now, because we get to sit at the big kids table at Thanksgiving, it's important for us to be able to work together with the office of electricity, planning how best to use the grid. And it's important for us to relate, well, with energy efficiency opportunities because the internet of things has come to energy efficiency in a big way. So information has impacted not just how we deploy solar, but it's impacted how we optimize what we do with our electricity. And roughly speaking, about half of the market for solar is on rooftops, commercial or residential buildings, and half of it is in big power plants. And in response to the need to build some bridges between different parts of the department, the grid modernization initiative was launched. Grid modernization is happening with or without the Department of Energy. It's happening all around the world. And so to work to bring these moving parts together, the DOE created the grid modernization initiative. A lot of different kinds of stakeholders are engaged here in a wide variety of ways, not the least of which is to come to consensus on the standards that are important for putting all these pieces together and have them be able to shake hands with each other and be able to get sort of VHS versus Betamax time, but with many different stakeholders who want their technique to be the standard. So being able to put these elements together as part of GMI as it's referenced and 13 of our 17 national labs participate in this effort called the Grid Modernization Laboratory Consortia. There won't be an acronym quiz at the end of the talk here, but there's a lot of acronyms that I've had to become familiar with. The important point I want to get across is that the grid is very complex and interrelated. And so these levels of the ecosystem from the in use to the generation all have financial dimensions that are relevant and stakeholders in the marketplace who now can make independent decisions without the need for a public utility commission to say whether or not they can put solar on their roof. So the opportunity here is also a challenge because there's now a mix of loads on the grid putting an electric vehicle changes, transformer loads in that neighborhood. There are a lot of challenges that the grid operators face, especially the distribution grid operators. And much of where I've spent time, fortunately just done the hallway from there are the folks in the building's office, David Nemsto and Kamal Sawyer are folks that have been thinking a lot about reducing energy consumption in buildings and bringing together the possibilities for the generation of solar to be a part either in the residential sector or in community solar kinds of configurations that can work for commercial applications as well. And much of the emphasis has been on zero energy building so that we make the best use of our electrons which helps further lower the cost of solar. So teaming with the buildings group has been one of my first priorities at having arrived at DOE. And the internet of things, the internet of things, the global market of the installed base plotted here in billions from 20 billion to 75 billion by 2025 represent a large challenge both for being able to have commonality and harmonization across how these devices interact with each other but also to assure the cybersecurity of our systems that are now so thoroughly networked and increasingly so being able to utilize in the best way possible the energy consumption in buildings is an important big step for what solar does on the distribution side of the grid. Well, so of course are on the transmission side. And one of the, I like this because it's the most simple graphic I could find that describes work that's going on at NREL that integrates energy efficiency with full of voltaics, with storage. Just having the modeling tools that can allow a planner, an architect, a building designer to go online and understand the trade-offs. When I got to DOE and spoke with the folks in buildings, they said the rule is, E-E then R-E. Energy efficiency, do all the things you can possibly do with energy efficiency, then put renewables on the building. And I said, well, you know, PV costs have fallen so fast that maybe putting in that perfect wall is more pricey than just sticking solar on the roof. So we've spent time looking at E-E plus R-E in our DOE sister group within E-E R-E. You've missed a wonderful world of evolution, I rule. And so we're now at E-E plus R-E for 95% of the organization. Being able to put those pieces together in an effective way is an important part of scaling the market for solar, which, as I mentioned earlier, is the driver for getting cost reduction. So when you visit the National Renewable Energy Lab, you can visit the Energy System Integration Facility the latest, greatest building at NREL. There's a 3D virtual reality space where you can put on goggles and you can see the voltage phase of power at various places on the grid and be able to optimize what happens with all those appliances or the heating ventilation air conditioning system with PV and with storage. Power electronics is playing an increasingly important role as we go from silicon-based devices to MOSFETs that are built off of silicon carbide, ultimately maybe into gallium nitride or beta-gallium oxide in the long term. But the rate of change of power electronics and the raw materials fundamental to making those devices is moving at a rapid clip and enabling the intelligence to go into the grid in the distribution part of the system and be able to help the grid operator, especially the distribution grid operator, manage the system. So interoperability has become a key part of how we work together in the different groups at DOE to find those standards. Here in California, Title 24 is a battleground these days where the development of codes and standards that bring PV together with energy efficiency are being debated by stakeholders who have a vested interest in better windows or stakeholders who have an interested in full-voltaics, the sizzle that helps sell a lot of the stakes that are related to energy efficiency. Title 24 of the code will issue in January of 2000. So here we are. There's three years of discussions and we're reaching a stage where the world is changing faster than government changes and a lot of the codes that we work on are evolving. The distribution grid, much of the grid itself has been in place for a long time. It also needs maintenance and upkeep. There are many opportunities for putting solar in and help extend the life of the grid. To keep things in perspective, solar is still a very small fraction of the capacity to generate power or the actual power produced on the grid. Still, we see that yes, continue to be a large part of our generation mix and capacity. This past year, there was more capacity added with solar than any of the other sources and it makes us even more important for us to work with the right sets of stakeholders for developing how to integrate these systems. And on the grid, there are some systems that have been in place over 40 years that are coming offline as these new sources of generation are being added. There's a tug of war around can we extend the life of those existing assets, do more with them and be able to minimize how much more new generation is needed on the grid. I sort of run through this quickly in the interest of time, but over here is a chart which basically shows the cumulative installations. China is now ahead of the rest of the world having over 77 gigawatts. There's a total of about 300 gigawatts now globally in solar generation and the expectations are that growth rate will continue to expand as time goes by, part of it here in the US and a lot of it internationally. So we have some great opportunities for being able to continue by reducing the cost of solar, innovating at both the solar cell end as well as the inverter end of the spectrum, being able to put the pieces together so that smart grids are everyday occurrence and in order to prove out the viability over long periods of time, we have a program that we carry out. Stanford and Slack are key parts of what we call Duramat, the durable materials part of what we do at DOE looking at what's been in the field, how well it's performed, what does that inform us about what we need to do to improve the design of modules and be able to have them last even longer with slower rates of change out in the field will make it even easier to finance these projects and to have policies put in place that sort of reinforce the value that we bring to the grid. And with that, I want to wrap things up. I thank you all for joining up today and for your attention here. Thank you.