Tesla battery packs: high-margin, non-automotive applications

Elon Musk has indicated that about 30% of the Gigafactory's output will go to stationary energy storage applications. I believe that diversifying product offerings to address multiple markets have many benefits to Tesla and love the idea of disrupting the utility market with distributed power, the combination of solar and batteries at point of use. At the same time, when you address new markets it makes sense to go after the highest margin applications first. As we've seen with Tesla's automotive products, the high margin products stimulate innovation and generate cash flow needed to bring next generation products to market. So in this thread, I would like to open up discussion on what high margin applications beyond automotive Tesla might consider.


In this discussion, I'd also like to think of "stationary" more broadly as non-automotive and there is a deep economic rationale for doing so. Batteries transform power at one point in time and space into available power at another time and place. Strictly speaking a stationary application is more limited because this translation only occurs over time and not space; whereas, mobile applications translate power over both time and space. At fundamental level, this implies higher economic potential in mobile applications than strictly stationary applications.


We see this distinction play out in the two basic energy markets: electricity and transportation fuels. Transportation fuels are dominated by oil, and the price of oil in this country is about 3 to 4 times the price of electricity on a power output basis. All fuels and renewable energy compete within the electricity market to deliver both power and energy at the lowest possible price. Oil does not factor large in power generation, only a few percent of electrical energy delivered, because it is too expensive. For the most part, the electricity market is a stationary market. It is very effective for things that can be plugged in while they run. But the advantage that transportation fuels have is that they are highly mobile and highly dispatchable. So the main contribution that oil makes to the electricity market tends to be to supply remote or backup power. For example, a copper miner operating in an area not served well by a grid may use diesel generators to power their operation even though this can cost more than $0.40 per kWh. The important thing to keep in mind here is that oil earns a very large premium over the electricity market price because it is mobile and dispatchable, that is it transfers energy over both space and time.


So when we consider the economic value of batteries, we note that they compete most directly with oil translating not just energy but power too over both space and time. Battery application will create the highest economic value where they offer both mobile and dispatchable benefits. These plumb applications will tend to follow oil. Firstly, EVs take cheap power from the electricity market directly into competition with the transportation market. EV owners benefit from this arbitrage opportunity to cut their fuel cost by 70% or more. This is this is primarily a mobility advantage that batteries afford, but there is a dispatchability advantage also. Batteries and electric motors are able to deliver power more responsively than an ICE can transform fuel into power. This is why a certain car that can go 0-60 in 3.2 seconds can be bought for a mere $105k. EVs can deliver superior performance because high power batteries are highly dispatchable. Both power and energy efficiency make EVs an attractive product and Tesla will soon be earning a 30% gross margin because of this.


So what comes next? What other applications might yield Tesla the highest economic returns based on both the dispatchability and mobility of it battery packs? Consider the remote copper miner again. The combination of both solar and batteries are able to deliver remote power at 70% below the cost of diesel power. Consider that the solar panels can produce intermittent power for say $0.10 per kWh. This saves the miner $0.40 per kWh while the sun shines. But their may be surges in power demand and sundown power needs that solar alone cannot address. A 1 MWh battery pack from fills this gap. Per charge/discharge cycle the pack takes in about $100 worth of energy and discharges $400 worth of power. This is a gain of about $300 per cycle. If this pack can deliver 1000 to 2000 cycles. Then it delivers $300k to $600k economic value. So Tesla could potentially sell such a pack for as much as $300 per kWh. Thus, Tesla might enjoy a 50% margin in such an application. What is critical to understand here, though, is that such a high margin is only possible because Tesla would be competing with the transportation energy market, not an efficient power grid. In the presence of a power grid that can hold the price to $0.20/kWh, then the pack only gains $100 per cycle, or $100k to $200k over its useful life. Thus, the price would need to be around $100/kWh to be competitive. In such a situation, the mine operator may value back up power or high power on demand (peak shaving) and thereby justify paying a higher price than $100/kWh, but the moral of the lesson should be clear: there is a lot more margin to be made competing with oil than with an efficient power grid.


So let's put on our thinking caps and see where non-automotive energy storage can yield the highest margin applications. I have a few ideas in mind, but I'd like to see what we can come up with collectively. I'd encourage us to take an open-minded brainstorming approach to this. Often ideas that seem far fetched at first can crack open unexpected possibilities.

Electric rail cars

Trains are already one of the most effcient ways to move freight, but are there any applications here for Tesla? Hybrid engines are being developed, but I don't think Tesla wants to get into battery electric engines anytime soon. There is another way, however, to add a battery boost to trains without designing new engines: self-propelled electric rail cars, SPERC.

Here's how it could work. Design or retrofit a rail car with a battery pack, electric motor and computer controller. Install a master controller in the engine or switch yard that communicates with each rail car controller. The master controller signals how much to accelerate or to brake.

In the switchyard, spercs can speed up the process of decoupling and coupling into trains as they move themselves into position. There is no need for an engine to switch cars from one train to another. This saves time, labor, energy and capital.

Trains with spercs reqire fewer engines and less fuel. Spercs aid acceleration of the entire train and absorb waste energy through regenerative braking. This is most advantageous in mountainous areas and populated areas where for safety reasons trains must frequently slow down. This could save time, energy and capital.

To put a rough value on it, suppose that every 6 to 8 kWh offsets 1 gallon of diesel. Thus, $0.60 to $1.00 of electricity offsets $3 to $4 of fuel. Let's say its a 75% energy cost savings. A 100 kWh pack on a sperc could save close to $30 per charge/discharge cycle. Over a life of 1000 to 2000 cycles, this is worth $30k to $60k. This supports a price of $300/kWh. The other time, labor and capital savings could make this quite attractive.

Moreover, the system is completely modular. Once the master controllers and charging infrastructure are in place, one can add any number of sperc to a train.

In this application, Tesla is able to add more value than just batteries. Its exertise in total electric drivetrain design, autopilot and digital control, energy pack management, and supercharging all come into play. Autonomous rail car technology could be really exciting. A Tesla begins to ship battery packs from Sparks to Fremont, I hope they start to play with the sperc concept. When all those battery packs are loaded onto a rail car, how can you not imagine what it would be like to add an electric motor to the ensemble? When you've got to move packs from one end of the Gigafactory to the other, how can you not think about how nice it could be to have a rail car that moves itself? The system pays for itself in energy savings, but it opens up a lot more potential than that. It plays to Tesla's strengths and solves real problems they face.

AudubonB, thanks for sharing your experience. It's hard for many of us to imagine what it is like to live in an area not well served by reliable, efficient grid, but in a global perspective these conditions are quite widespread. Something like 20% of India is without grid access and in many other area the grid is quite unreliable. So solar + batteries is quite common. SolarCity has set up a non-for-profit foundation to electrify schools in remote villages that are not served by utilities. These systems are solar +battery and support economic development through fascilitating education, community groups, pumping water and providing a place.to recharge phones and other devices. Ironically there are many areas where wireless phones have leapfrogged ahead of both landline cables and electrical grids. You can get wireless service but charging your device is a major challenge. Like wireless, PV + batteries is a distributed technology that can leapfrog well past outer reaches of efficient grids. It is really hard to fathom what kind of economic development this kind of technology can facilitate. One generation grew up without any electricity and the next generation is able to surf the internet and take university courses online.

So what is the opportunity for Tesla? Clearly offering cheap batteries is one thing, but this is likely to be pretty low margin. How does Tesla use more of its innovative capacity to deliver a higher margin product? Demand Logic moves in this direction. It is more than just a battery. It is an intelligent energy storage system with computing power to optimize the economic output of the system. Basically, it can be programmed to optimize your power bill. It can also communicate with other devices and entities to achieve higher system order efficiencies. For example, suppose SolarCity has 1000 10kW units in a service area. There is the potential for SolarCity to coordinate 10MW of power and negotiate better realtime rates for all its users with the wholesale power market. Tapping into the volatility of the wholesale market opens up greater arbitrage potential than what a residential rate payer generally has. So this is the right sort of direction to go within the context of an efficient grid.

But let's return to the challenge of providing power in remote areas. The programmability of Demand Logic may open up some helpful functionality for microgrids. Suppose the businesses in AudubonB's communities wanted to reconnect into a new microgrid. Suppose the each had a few power generation devices such as solar panels and diesel generators. We would want Demand Logic to integrate these for each user. So Audubon should not have to fire up the generator when needed, rather Demand Logic would do that for him. This is the first level of integration, single user. The next level is connecting with other user. Suppose that neighbors had their own Demand Logic device or something compatible. How could you get the devices to communicate with each other and share power? Moreover, how could the costs and benefits of such a network be shared equitably? A free market approach may just work. Suppose AudubonB's Demand Logic device is programmed to buy and sell power for him. It knows how much energy it has in storage, how much power the solar panels are producing, how much his lodge is using, and how much it costs to run his generator. Taking all that into consideration and the long-term costs of storage and diesel generation, it computes computes ask and bid prices. His neighbor's DL devices do the same and execute realtime trades of power. So occasionally he might hear his generator fire up, not because his lodge needs it, but because his neighbor needs the extra power. Suppose in a few years his generator stops working. This energy trading scheme might just make it so he won't have to repair or replace it. DL simply buys power from neighbors as needed. Thus, the capital cost for all users in this microgrid goes down over time and reliability goes up.

In keeping with my earlier theme, a big chunk of the arbitrage value in this microgrid is competing with the price of diesel. With more sharing, Audubon may be able to put more money into PV than diesel generators and get a better return. This would be true for the whole microgrid.

Right. Full concept approach is the way to go. It reminds me of Apple's plug and play approach. These things need to integrate to achieve results that merit high margins.

To get to 400 GFs in twenty year, we need to double capacity every two years or so. This is comparable to the way solar is currently growing. So it could be done. But I don't think that a race to commodity prices and razor thin margins is the way to go. A small, highly innovative company can focus on high margin applications, even if the scale is not apparently massive.

Oncor wants to pay $200 per kWh for distributed grid storage in Texas and get compensated $150 per kWh for installation and maintenace. I think the likes of SolarCity could install for much less. So Oncor wants to be the system integrator and get low cost batteries from Tesla. If the value of this grid storage is really $350, are there better ways for Tesla to get more than just a $200 slice of that pie? It sound like Oncor, a transmission and distribution (T&D) operator is simply trying to protect itself from disruption from distributed energy storage. SolarCity can install storage directly in homes and businesses for less than $350 per kWh. In fact, they have recently been shopped and offer a 10 year lease with present value of about $270 per kWh. T&D utilities will have to compete with that, so it makes sense that they might try to preempt this and lock in costs that they can capture over time through regulated rates.

Thanks, AudubonB, Good to know. I am by nature a highly conceptual thinker, so it is always helpful for me to get more gritty information from experiece to round things out. So the question I have for you is what sort of innovation could Tesla bring to market that would most help Paxson? Tesla's Demand Logic product is programmable and networkable, so the question is how to harness this capability to meet the needs of a community like Paxson. I'm thinking that with just a little more hardware, the whole thing boils down to a programming problem.

By analogy, think of where Tesla is with Autopilot. They start with a base car that is almost entirely digitally controlled. It's already highly programmable and networkable. They add digital brakes and a battery of sensors and turn programmers loose on coding Autopilot functionality. They've got the digital car platform, so adding more autopilot functionality is just a matter of programming. Moreover, consumers are ready to buy these products now under the expectation that software upgrades will only enhance these vehicles.

So taking this analogy back to stationary storage, what hardware needs to be added to the system so that you can network power sharing? Get that in place and then the rest of the solution is a matter of programming. If I were Musk, I think I would want to send a team of engineers and developers to Paxson to see what it would take. My hunch is that if they can innovate a solution for Paxson, they can build on that and deploy all over the electric frontier. Not only is this potentially an exciting product, but expanding the electric frontier is important to long term EV adoption.

Thanks, Robert, for the information rich reply. For me, your analysis shows two basic problems with long-haul shipping, and it's instructive for other applications as well.

First, the energy arbitrage potential is weak here. Apparently, bunker fuel is pretty cheap so that the ration of fuel cost to electricity is pretty low, just 2.25 to 1. In other high margin application, we have considered ratios closer to 4 to 1, which gives you a 300% gain on energy. But in the shipping case we are looking at only a 125% gain which is going to make it hard for the battery to pay for itself over its life time. So this is clearly not the sort of high margin opportunity I am looking for.

The next problem is slow cycling. Suppose the energy arbitrage was in fact high, a 4 to 1 ratio leading to a $3M savings per 21 day voyage. If the batteries cost say $3B, the vessel would need to make about 1000 voyages to break even, but cycling batteries every 21 days means it would take over 57 years to break even. This is just not acceptable. In 30 years, the price per kWh could come down to $25, which would break even in about 7 years, and this might pencil out at that time. But the lesson to be learned here is that cycle times must be short, like daily, to get enough cycles within a reasonable payback period. If this ship was able to recharge daily, it would only need a 700 MWh pack and could achieve 1000 cycles within 3 years. So if the energy arbitrage was high enough or the price of battery low enough, it could break even in reasonable time. The cycle time needs to be short.

I've got such a marine application in mind, which I will write up in a day or too. Bonus points to anyone who can guess what it is.