Emerging storage business models: Part II
Storage is coming down rapidly in cost, and developers are figuring out ways to tap new revenue streams. A group on the front lines of the storage business talked at an Infocast conference in San Francisco in late February about the evolving storage business models.
The panelists are Karen Butterfield, chief commercial officer of US storage company Stem, Raphael Declercq, vice president of portfolio strategy for EDF Renewable Energy, the North American arm of Electricité de France, Sam Jaffe, managing director of Cairn Energy Research Advisors, an energy storage consulting and research firm, and John Jung, CEO of Greensmith, the US storage arm of Finnish company Wärtsilä. The moderator is Keith Martin with Norton Rose Fulbright in Washington.
Utility-scale storage
MR. MARTIN: One thing new industries must do is find the right business model to get traction.
The solar rooftop industry took off when SolarCity and others pioneered a third-party ownership model where the solar company put solar panels on customer roofs for free. The customers signed 20-year contracts to buy electricity or lease the solar systems.
What business models are taking shape in the electricity storage business?
Let’s start with utility-scale storage. There seem to be four main business models.
One is where a standalone battery is bid into an organized market, like PJM. The battery is offered each hour to provide frequency regulation services at whatever price is established by auction that hour. The battery owner receives a payment from the grid. If the auction price is $25 a megawatt hour and the battery owner bid 20 megawatts, the battery owner receives $25 times 20 from the grid for the ability to use the battery than hour.
Another utility-scale business model is a tolling agreement where the battery owner stores electricity for the local utility for a fee. The fee may be a variable hourly fee like an energy payment tied to how much actual use there is of the battery. It could be a fixed capacity payment that is like a reservation charge for the right to use the battery that hour. It could be a combination of the two. This is the business model used for the 110-megawatt battery that is part of the AES Southland project in southern California.
The third model is a buy-sell model where the battery owner buys electricity during off-peak periods when the electricity is cheap and sells it back to the grid during peak hours. This model focuses on time-based arbitrage. It most common use is in pilot-scale storage projects.
The fourth business model is where a large battery is added to a solar or wind project. It regulates the ramp rate at which electricity from the project is fed into the grid. It also puts the project in a position to earn additional revenue by providing ancillary services.
Are there other utility-scale business models that are not on this list?
MR. JUNG: The way you get value out of energy storage, solve problems and make money varies depending on not only who you are, but also where you are.
For example, we have an 80-megawatt hour system that is doing four things to make money for the customer, AltaGas. It is doing resource adequacy, which is a four-hour product, for California. It is also making money by providing frequency regulation services. It is making money in the day-ahead power market. It is also making money in the five-minute market. At some moments, the price in the five-minute market exceeds $1,000 a megawatt hour.
The key to succeeding at storage is to combine as many different applications as possible. Your ability to do that depends on where you are and who you are.
MR. MARTIN: This is what is called value stacking. Your point is that it is not possible to draw clear lines around four current business models in the utility-scale market? Each of the models has elements of value stacking?
MR. JUNG: No, you can. My main point is that business models mean different things in terms of how customers make money and how the storage company makes money.
For example, we had two 10-megawatt systems going in ERCOT recently. The way in which the business model monetizes systems in ERCOT is very different than in California, PJM, New York or other places.
MR. MARTIN: The question is whether there are other basic business models for utility-scale batteries than the four I described.
MR. DECLERCQ: Another model is to use a battery as a way to avoid investment in transmission or distribution. This is something that we have done at the distribution level for smaller utilities in the northeast where we compare the cost of adding poles and lines to support additional load at the end of the line to putting a battery there instead.
Another comment is you could combine your third and fourth business models. Energy arbitrage with a standalone battery is not economical today; but adding storage to solar is starting to be economic at utility scale, especially if you are not only focused on your internal rate of return, but also on reducing your risk. A solar project with a battery is more likely to be in the money on any hedge.
MR. JAFFE: Let me add to your list my pet favorite, which is the new First Solar-Arizona Public Service announcement. It is essentially a solar storage peaking plant.
MR. MARTIN: Describe how it works.
MR. JAFFE: It is a solar park plus energy storage that is used solely to provide peaking capabilities in the middle of the day. Traditionally, utilities spend 85% of their resources on the 15% higher part of the peak. That has always been the most challenging part of managing the grid. If you can address that challenge with solar plus storage rather than gas peakers, that is potentially a very significant development for the grid.
MR. MARTIN: Why do you need to add storage to solar to provide peaking capability during the middle of the day when solar is at maximum output?
MR. JAFFE: Cloudy days. If solar is offered as the solution to serving peak load, then you have to guarantee you can solve that problem every day and, even if you are in Arizona where there are two days a year with clouds, you have to have some sort of solution for those two days.
Distributed storage
MR. MARTIN: Let’s move to behind-the-meter models.
One business model is a storage company installs batteries, retains ownership and charges customers either a subscription fee or a percentage of the customer’s energy savings. The storage company manages the battery to reduce the amount of electricity the customer draws from the grid. It uses software to predict how much the battery will be used and when there will be spare capacity. Karen Butterfield, what is a typical subscription fee under this model?
MS. BUTTERFIELD: It is calculated the same way a PPA price is calculated. It is cost based. We gravitate toward markets where the customer can reduce its utility bill by at least 200% of the fee charged.
This is just a rule of thumb that we use. We have found over the last year that many customers are willing to take less than that, especially in the public sector where customers are very interested in sustainability initiatives and in promoting storage.
MR. MARTIN: So you say to a potential customer, “We will put a battery on your premises. We will manage it with software to try to manage your use of energy so that you save money.” You charge a subscription fee. It is a periodic fee, and you set it at a level that ensures the customer is getting at least twice the savings as the fee.
MS. BUTTERFIELD: That is a good description.
MR. MARTIN: What is a typical fee for a business? I think you are focused on putting storage on commercial properties.
MS. BUTTERFIELD: We install systems anywhere from 100 kilowatts to two megawatts in size, so the cost and the fee vary depending on the energy savings and the cost of the equipment.
The important part is we go to markets where we can convince the customer to think, “Why wouldn’t I do this? There are savings for me.” Then the real beauty of the model is that as new programs come along, we will go back to the same customer and say, “We can bid you into something called DRAM in California — demand response auction mechanism — or we can use the battery to bid into a demand-response program in Hawaii.
We have a system sitting there at the building that is generating savings for the customer. We go back to the customer and add another value stream. We might modify our service fee and increase their savings, or give them a cut of the additional revenue. As the regulatory world changes and more value streams become available, we can share the benefits with our customers.
MR. MARTIN: It is a little like cable television. You keep adding more channels.
MS. BUTTERFIELD: Yes, the sports package is extra. [Laughter]
MR. MARTIN: Sometimes you charge the customer not a monthly subscription fee, but a percentage of the savings. What percentage of savings would you typically charge?
MS. BUTTERFIELD: Sometimes that model is not as easy to finance and so if we do not have to go in that direction, we don’t. We have some markets that are opening up where that seems to be the flavor of the day, and we are certainly going to participate in those markets.
MR. MARTIN: Let me ask you three more questions. How long are the contracts typically with the customers?
MS. BUTTERFIELD: Ten years.
MR. MARTIN: Does the customer have to buy out the back end of the contract if he or she cancels?
MS. BUTTERFIELD: Yes, like a solar rooftop power purchase agreement.
MR. MARTIN: Why are you focused solely on commercial and not also residential?
MS. BUTTERFIELD: Because of the rate structures and the software and equipment that you have to put at the site. Most residential customers do not face demand charges. But for commercial and public customers that do face them — they can be more than 50% of their bills — we offer automated savings using artificial intelligence. That captures data for the customer at one-second intervals, provides real-time metering, and uses five-minute-or-less dispatch response without their involvement. That is the primary reason today.
MR. MARTIN: Raphael Declercq, EDF owns groSolar, which is also in this business. Does its business model work the same way as Stem’s?
MR. DECLERCQ: Partly. We have more flexibility on the financing side because we can do it on the balance sheet, at least for now. So we can do some shared savings where the customer gets something for nothing. The customer gets a share of the savings that the battery is going to generate. We compare the electricity bill as it is today to what it would have been if the battery had not been installed, and we give something between 15% to 40% of the savings to the customer.
You have to take into account some behavioral economics, too. If you agree to pay the customers something, then the customers may be willing to pay larger subscription fees. That helps with financing.
In some cases, we have moved to lease payments. They are a fixed amount each month.
MR. MARTIN: Are you finding customers prefer leasing the batteries?
MR. DECLERCQ: It depends on the customer.
MR. MARTIN: How long is the lease?
MR. DECLERCQ: Typically 10 years.
MR. MARTIN: Sam Jaffe, you wanted to mention something.
MR. JAFFE: It is important to understand we are talking about two different business models. There is the business model being used with the customer. But then Stem or EDF or groSolar is not only providing a service to that customer, it is also making revenue from aggregating all the customer batteries to turn them into a virtual power plant that I assume is almost always tied back into some form of contract with the local utility or grid operator to sell it unused storage capacity.
Would either of your firms be able to do just a customer model without that aggregation or virtual power plant?
MR. MARTIN: Before you answer, let’s stipulate that the second distributed business model is where you offer the spare capacity on the customer batteries to the local utility.
MS. BUTTERFIELD: In some markets, it is economically viable to work just with the customer. However, as a venture-backed company, our investors are interested in seeing us build the largest storage network possible as quickly as possible. They want the additional value streams offered by a virtual power plant.
In some cases, we start with the utility contract, but most of the time, we find places where the customer economics work well enough to build a fleet of batteries, and then we try to add that other revenue stream.
The AI software to manage everything is key. We call it Athena. We have data coming every second from the building load. We have a price coming from the market. We have hourly temperatures coming in from weather services. All of this data must be managed in real time.
Every single building, every single market, every single tariff has to be managed. Athena is able to find the optimization point where Stem can save more money for the customer when a utility says, “We will pay you for a demand-response rate.” That is really the future.
The reason we are selling so much so fast right now is costs are coming down, customers are comfortable with the technology, and the customer-facing business model is simple.
MR. MARTIN: This is big data. It is artificial intelligence. It is a software business. You are managing these assets with the help of software to optimize their use. Are the software engineers or the salesmen at the top of the pecking order at a storage company?
MS. BUTTERFIELD: I run the sales organization, so . . . .
MR. MARTIN: It is the sales people. [Laughter]
MS. BUTTERFIELD: I am kidding, of course. We work together as a team. When something changes in the marketplace, we all change. We work with the product managers, the development team, and we say, “We know we told you we wanted to develop these algorithms for this market, but this is happening faster.” We do a quarterly planning process.
The sales people have their ears to the ground. They get ahead of the technology people. Then you have tech debt, and they have to chip away at that tech debt. Then the sales people have to catch up. It is like a see-saw. The two roles are equally important.
MR. MARTIN: How does it work for the customer? The customer has a battery that is being used to manage its energy usage, but you are also offering the local utility the right to use the battery. Does the customer have first claim on the storage capacity?
MS. BUTTERFIELD: Not exactly. We have made a commitment to the customer to save it a certain amount each month on its utility bill. If we are also making this optimization decision about how to get more grid revenue, which the customer partakes in, or how to get more demand-charge savings, which the customer partakes in, or how to get more out of the demand-response program, which the customer partakes in, the customer should be fine with use of the battery in that manner.
MR. JUNG: Can I offer a contrasting picture? On the one hand, you have a kind of SolarCity no-money-down type of model. There are a lot of interesting aspects to that model, especially in so-called behind the meter-type applications where, for the most part, I think the use case has largely started with demand-charge management.
On the utility side or the grid-scale side, which is where we have dwelled, it was very simple. Big companies like NextEra or Duke or E.On or AEP want to put the storage system into rate base. If the customer is an independent power producer, then it wants to own and operate the assets like any other technology.
The difference — and I want to amplify the software aspect — is that while these utility-scale customers are buying a piece of equipment, the thing that is new to these very large power companies is they are now also getting a software license.
In some ways, the utility-scale model is a lot simpler because we do not have to own and operate. The big power companies can use their own balance sheets. When you can tell someone that the return on investment is not just the denominator in terms of waiting for prices to come down, but also the more value streams you are able to capture. It leads to a better outcome.
This software thing is pretty important. We are already on our sixth generation of software.
When you talk about the multiple things that energy storage can do, you need software in order to be able to do them.
MR. MARTIN: This is as much a software business as a hardware business. Raphael Declercq, you were about to say something.
MR. DECLERCQ: Both the commercial efforts and the software development are expensive. This is something that an investor should look into because making these business models sustainable in the long term is not an easy thing.
MR. MARTIN: Part of your selling point is that you are offering software to manage the battery.
MS. BUTTERFIELD: Yes, but we call it customer acquisition costs. If you sell two things, you are better off than just selling one. If you can sell three things, you are better off still. You look at opportunities. One thing we did recently is we partnered with CPower on the demand-response side, and now we can take the sales organization and multiply the coverage. We can do demand response and storage at the same site of the customer. This helps to reduce your customer acquisition costs as a percentage of revenue.
Customer acquisition costs
MR. MARTIN: One of the challenges of the residential solar rooftop industry is the high customer acquisition cost. It is about 25% of an installed system. What is it for storage?
MS. BUTTERFIELD: It is nowhere near that, and it is nowhere near that for commercial solar either. But it is still substantial. When we were selling 30-kilowatt systems, our customer acquisition costs were off the charts. Now we are selling one- and two-megawatt systems to large universities and the customer acquisition costs are a smaller fraction of that, but still meaningful.
MR. JUNG: The economies of scale are really important. The 20-megawatt, 80-megawatt-hour system can be installed in about four months. It takes the same effort as to install a one-megawatt, four-megawatt-hour system. The bigger, the better.
If you take a look at the supply-chain cost or at the total cost of ownership of, let’s say, solar which I think people understand a lot better than energy storage, it is kind of similar. Most contemporary studies show that the total cost of ownership at the residential level is two to three times higher than the utility level. Why? Because it is just a ratio of how many installations, how many points of failure you need to manage, and just the ability to buy 100 megawatts versus five kilowatts of solar panels and inverters. It is all that kind of supply chain stuff.
The cost of energy storage is falling because the cost of lithium ion has fallen by 50% in the last 18 months.
We buy a lot of batteries around the world. The cost is also falling because each successive installation of energy storage is getting faster.
The counterpoint is many energy storage companies have gone out of business. It is hard to tell which one will be the Uber or Google and which ones will fail.
MR. MARTIN: This is a typical pattern in any new industry.
MR. JUNG: We published a white paper called “Futureproofing Energy Storage” because the average tenure of most of these contemporary systems, although the warranties are 10+ years, is actually about three years. People do not know the eventual shape of the degradation curve and what will actually happen 10 years down the road.
Potential revenue streams
MR. MARTIN: That’s a lot to chew over.
Let me wrap up the distributed business models. We have 10-year leases of batteries. We have a service model where the storage company charges a subscription fee or a percentage of savings. Within that model, companies are also offering the spare storage capacity to the local utility to earn more revenue.
Two other models are direct sales of batteries to homeowners, and then there are solar rooftop companies that are installing batteries in connection with rooftop systems and charging for their use.
All of you have talked about trying to add more revenue streams. The Rocky Mountain Institute says there are as many as 13 potential revenue streams. How many is the industry realizing on today and which ones?
MS. BUTTERFIELD: It is probably on the order of four or five.
MR. MARTIN: What are they?
MS. BUTTERFIELD: Demand-charge management. Solar plus storage, so that would be ramp rate. There are plenty of people doing ramp rate.
MR. JUNG: Frequency regulation service is probably there too, Karen?
MS. BUTTERFIELD: Behind the meter?
MR. JUNG: Yes.
MS. BUTTERFIELD: We are starting to see it in one or two markets.
MR. DECLERCQ: There are some demand-response programs, too. I don’t know if they were included in demand-charge management.
MS. BUTTERFIELD: I think a non-wires alternative is one of the other items.
MR. JUNG: It depends whether you are behind or in front of the meter. On the grid-scale side, in Texas for instance, the energy storage systems are being fed in Roscoe entirely by wind, and they are addressing capacity. ERCOT looks at fast responding frequency as a separate product.
In California, the applications are multifaceted because a big system does four different things. To be clear, these use cases are not just expanding beyond the four or so that we mentioned, but the phenomenon of value stacking in multiple markets is also happening depending on what part of the world you are in.
We just delivered a German system that does multiple things. We just got awarded a system in Hungary that is dealing with frequency. That is its day job. It does other things on the side.
MR. MARTIN: Stop on the German system. It does multiple things. What are they?
MR. JUNG: Number one is it helps the grid manage frequency. Germany has periods when power suppliers have to pay the grid to take their electricity because Germany has a lot of solar. The storage system can do some peak shifting.
MR. JAFFE: And also capacity.
MR. JUNG: If you are a developer who has put a lot of solar in, the worst word you can hear is curtailment. You just spent $100 million on a system that you are only allowed to use 70% of the time. More markets are combining these use cases, which I think is a really positive thing.
MR. DECLERCQ: I am trying to think of other use cases still behind the meter. The contract we have with PG&E for 10 megawatts and 40 megawatt hours behind the meter serves PG&E with resource adequacy, and then we do peak shaving for the customer. So there are two revenue streams that are stacked.
The interesting thing about resource adequacy is that it is driven purely by the utility. It is an accounting matter. Sometimes the utility will do other things with the energy that we provide. If PG&E decides that it wants us to discharge at another time, for example, we may do so, but we don’t know exactly what the utility is doing with that energy. So there may be other hidden revenue streams that are in the hands of the utility.
MR. JAFFE: There is now a UK capacity market specifically for storage. But also in response to the Rocky Mountain Institute comment, our taxonomy is a little bit different. We have over 25 different profit models for over 25 different applications of energy storage, and I can say that eight of them are now in the money in various places in the world. Two years ago, there was only one.
MR. MARTIN: Twenty-five different models. Do they go beyond what we have discussed here?
MR. JAFFE: Yes. Essentially they are segregated more finely than the way that the Rocky Mountain Institute is looking at this. How do you own a battery and make money off of it? Any way we can think of.
MR. JUNG: If you look at the ancillary services market, frequency regulation service — one type of ancillary service — is done differently around the world, so there could be five or six different revenue streams there. From a software standpoint, we have seven different applications out of the box, but the algorithms underneath those applications, like for frequency, vary from one location to the next. If you treat each of those as a different revenue stream, that is about 30 application streams.
MR. JAFFE: We cover batteries for cars, too. Cars are easy. You put a battery in a car, you sell the car, it goes. Stationary storage is so complex and sophisticated and that is a sense of what this market really is. A few years ago, we didn’t know what energy storage was. Companies like these three are figuring it out as they go and starting to make profitable business models out of it.
MR. MARTIN: And testing different business models to see what sells in the market.
Arbitrage
MS. BUTTERFIELD: One topic we touched on is arbitrage. Many people think that that is the name of the game. You buy low, sell high. The spread is just not enough to make the model work, but the spread exists, and it is incremental. It is a value stream that we didn’t even mention. There are more and more places where that spread is worth chasing.
MR. MARTIN: The spread between what and what?
MS. BUTTERFIELD: Between the charging price and the discharging price of the battery. So you could charge at 8¢ a kilowatt hour and discharge at 30¢ a kilowatt hour. You are saving that customer the difference between the two rates.
MR. MARTIN: Is that being done currently in California?
MS. BUTTERFIELD: Yes. It does not stand on its own yet because the spread is not substantial enough in any market, but adding it to an existing model and using the Athena software helps to capture it. All of a sudden the utility changes those rates and the spread changes. You have to be able to react to that. That is what the software does.
MR. MARTIN: Stem aggregated the capacity reserves on lots of batteries at florist shops, grocery stores, and so on in southern California and sold them to Southern California Edison. What percentage of the revenue is coming from capacity and what percentage from subscription payments or energy savings?
MS. BUTTERFIELD: I would not be able to share that information in this public forum.
MR. MARTIN: Audience, go ask her after the session.
MR. DECLERCQ: There is another part of the equation that is not really a revenue stream, but that really matters in California, and that is the SGIP. I don’t think the economics work without SGIPs behind the meter.
MR. MARTIN: Explain the SGIP.
MR. DECLERCQ: It is the self-generator incentive program. It provides a subsidy spread over five years as an inducement to make sure a storage system actually works. It is a well-designed program as far as we are concerned.
MR. MARTIN: Do you have a revenue breakdown for how much is SGIP, how much is subscription payments, how much is capacity payments by utilities for the right to use the batteries?
MR. DECLERCQ: It depends on the particular case. What I can tell you is that SGIP is very important. In our experience, the installation does not work without it.
MR. MARTIN: 10%?
MR. DECLERCQ: It is more than that in the first five years. It is closer to 40% to 50%.
Potential issues
MR. MARTIN: Sam Jaffe, wear a different hat. You are now a lender or maybe an equity investor. Where would you probe before investing in a storage project?
MR. JAFFE: Degradation. Former US defense secretary Donald Rumsfeld said there are things I know I don’t know and things I don’t know I don’t know. In this sector, there is a lot of not knowing what we don’t know. We do not know how long the batteries will last. There are five or six ways you can manage that risk, but right now essentially what is happening is people are relying on the largest Asian conglomerates that make the batteries. Part of that is for quality, part is for expertise, but mostly it is for balance sheet. These guys can back up their own warrantees.
MR. MARTIN: So people will probe on degradation, but won’t find an answer.
MR. JAFFE: We will know in 10 years whether a battery lasts 10 years.
MR. MARTIN: That is not so comforting for a project that relies on a 10-year offtake contract. Raphael Declercq, where would you probe first as a lender or equity investor?
MR. DECLERCQ: I would look for a strong parental guarantee because, at this point, there are a lot of unknowns. I agree with Sam Jaffe. Our industry is still in its infancy. There are a lot of things, like rate of degradation, that we do not know. Degradation is the biggest risk.
I would look also at who is behind the project. It matters. This is like the early days of solar or wind.
MR. MARTIN: You want a big company like EDF?
MR. DECLERCQ: No, no. But seriously, the customer or lender who goes with a startup takes a high risk. Maybe they are aware of that and perhaps that is why there have not been many systems financed by third parties, except with very, very rich contracts. .
MR. MARTIN: Karen Butterfield?
MS. BUTTERFIELD: I would focus on the data. The important thing with batteries is how often you use them and how you use them, and when you charge them and discharge them. Our relationship with our financiers is premised on being able to provide them with data that shows what we are doing.
What happens when you want to change what you are doing? It has to work financially. So what if you burn the battery out in eight years if you have made 10 times the amount of money that you thought you were going to make? Just replace it after eight years.
MR. MARTIN: John Jung.
MR. JUNG: I think we are gathering more data about how these batteries operate in the field than even the OEMs are. So data is very valuable in and of itself, but I think also that nothing supplants experience. We have had a chance to integrate 16 different batteries since we started the company 10 years ago. We are all in the risk business in this room. That is how we make money. I like to say the serenity prayer, which is, “May I have the serenity to accept the things that I cannot change, the courage to change the things that I can, and the wisdom to know the difference between the two.”
MR. MARTIN: This is not encouraging for investors. [Laughter]
MR. JUNG: It is, actually. I can say it given that we have delivered seven times returns to investors. We need to make sure that we manage technology risk.
MR. MARTIN: Sam Jaffe, each of the three has basically just made a sales pitch for his or her own company. You are an unbiased advisor. Is there anything else you would add if you were a lender or investor? Where would you probe?
MR. JAFFE: Another point that was touched on earlier is the changing landscape of how an application works. If you are doing demand-charge mitigation behind the meter in San Diego, for example, what happens when that problem is solved and demand charges contract dramatically? All of a sudden you have a great return-on-investment model that just disappeared. Anticipating that and trying to understand future dynamics of regulations and markets is important too.
Editor’s note: “Part I” on this topic can be found in the April 2017 NewsWire.