I’ve written more than a few blogs about energy before. Why would I write another? I can’t think of a single reason.
More seriously, I recently reviewed a personal whitepaper I wrote in January 2019 about battery economics and feel it is now time for it to see the light of day. The reasons for my reticence about publishing at the time should be obvious enough by the end of the post!
No, I Will Not Shut Up About Solar And Batteries
But first, a quick review. Slices of insights have made it into other, intervening posts.
Most recently, I contributed a piece to The Cosmopolitan Globalists about the ascendancy of solar, wind, and batteries for the next century of electricity generation, particularly in contrast to legacy technologies that cannot hope to match solar’s pace of innovation and deployment.
The future of electricity is local generation and consumption. It is my considered opinion that reductions in the cost of primary electricity production due to innovation in solar panel manufacturing are already trickling down to the consumer, and continued reductions in cost will preferentially favor the development of local storage capacity over the long distance power transmission traditionally favored to pool demand in a predictable way.
I just checked the calendar and it’s been nearly a whole year since Australia managed to prostrate itself before the world as the paragon of terrible energy policy and look, right on schedule, they’ve done it again. Good work, guys!
The upshot is that energy systems are transforming the world over. The new electricity is resilient, cheap, clean, innovative, productive, and just damn cool. In particular, annual growth in battery production investment has hovered around 250% for a while and as a result we’re seeing meaningful penetration of grid scale battery storage systems beyond pilot plants, such as the South Australian Big Battery, which surprised everyone when it paid for itself in just three years. Well, nearly everyone.
Consider those numbers in the context of typical energy infrastructure. Australia has base load coal plants nearing half a century old that are now well and truly stranded assets. They’re already falling apart and have never paid for themselves beyond operating expenses. Energy infrastructure is vitally important but let’s not kid ourselves, it’s not the sort of industry where we traditionally saw more than a few percent of annual growth, free cash flow, or unsubsidized operations on any scale. Into this mix comes a new form of power so transformative that private markets can build and operate it, without subsidies, at a cost so low that the costs of transition can be absorbed with barely any inconvenience.
In a rapidly expanding set of markets, new build solar and batteries are cheaper, including the cost of construction, than continuing to operate legacy thermal electricity plants. The writing has been on the wall now for more than a decade. Yet some industry actors spent most of the 2010s convincing themselves that while burning coal was a thing of the past, burning fracked gas to generate electricity was immune to competition. Whoops.
What’s The Latest?
The latest news is that solar and batteries have just bagged another target. Gas peaker plants, already wounded by the Horndale battery, are now continually losing new build contracts to battery plants.
There have been a spate of recent articles. Here are two excerpts from the Wall Street Journal.
Even in Texas, a state with a fiercely competitive power market and no emissions mandates, scarcely any gas plants are under construction, while solar farms and batteries are growing fast. Companies are considering nearly 88,900 megawatts of solar, 23,860 megawatts of wind and 30,300 megawatts of battery storage capacity in the state, according to the Electric Reliability Council of Texas. By comparison, only 7,900 megawatts of new gas-fired capacity is under consideration.https://www.wsj.com/articles/batteries-challenge-natural-gas-elecric-power-generation-11620236583
I note here that Texas has probably the cheapest natural gas in the history of the universe. They routinely flare it! And even so, batteries are so much cheaper that they’re out-competing gas by a factor of five.
Prices have come down a long way since January 2010, when Boston Consulting Group estimated battery costs at between $1,000 and $1,200 per kilowatt-hour. It said getting to $250—a level car makers were targeting—“is unlikely to be achieved unless there is a major breakthrough in battery chemistry.”
Today, battery prices are about $125 per kilowatt-hour, after big increases in manufacturing capacity lowered costs, and tweaks to chemistry and design yielded further savings.
Battery costs are widely expected to fall further, said Venkat Viswanathan, an associate professor of mechanical engineering at Carnegie Mellon University. He expects them to go as low as $80 per kilowatt-hour in two to three years before bottoming out.https://www.wsj.com/articles/the-battery-is-ready-to-power-the-world-11612551578
If you are not partial to the taste of hat, do not bet that solar, wind, or battery price improvements will “bottom out” in the next few years. Or ever.
Since I know some readers are going to skip the articles and complain about supposed resource scarcity, the energy and materials required to build out these battery plants are, on a per capita basis, less harmful to the environment than comparable consumption of cars and trucks alone, including oil extraction and refining. It’s not even close. Roughly 100 kg of recyclable lithium per person per lifetime sounds like a lot, until you consider that on average people consume this much gasoline in three weeks, or water in just eight hours.
Okay, we know what is happening. This sudden and rapid transition came as a surprise to even veteran Green activists. Why is capitalism suddenly the good guy here? Why is economics achieving in just a few years what decades of achingly slow international negotiations on emissions control largely failed to achieve? You know, aside from alignment of incentives and the fact that no government is now required to unilaterally deindustrialize and impoverish vast swaths of their population. Are you listening, Canberra?
Why are batteries so unstoppable? Why are they printing money as fast as they can be deployed? Why is the transition to storage-by-default so destabilizing to traditional electricity business models? What are the limits to arbitrage in this case?
Let’s State The Obvious
Batteries can be used to both stabilize the grid, make money, and aggressively drive out gas competition.
Wait, Is This Actually Obvious?
There is a non-obvious true fact here. Usually the addition of a successfully integrated component to a system reduces, not induces, demand for more of the same. In the case of batteries, this is not the case. Like Starlink, the beachhead minimum viable product only induces more demand for the same system, methodically displacing almost all mainstream competition.
Common, Incorrect Belief About The Universe
The common belief is that there is only one way that the electricity market can function, where electricity is produced at exactly the same rate that it is consumed and transmitted instantaneously from generators to customers via wires. Indeed, grids use generation frequency as a feedback mechanism to regulate production, while relying on large numbers of customers to smooth out demand variation and improve predictability of, mostly, diurnal power demand cycles.
In fact, there are multiple competing mechanisms for grid operation, though physical reality is obfuscated by varying regulatory regimes as well as legacy hardware and accounting systems. In this case, it turns out that sufficiently cheap batteries combined with cheap albeit intermittent solar power is the secret sauce necessary to destabilize the previously cozy “stable attractor” embodied in traditional grids. Not only is solar+batteries cheaper, it doesn’t necessarily play nicely with legacy systems and, in fact, can actively compete with them on price. Due to the versatility of sub millisecond power control enjoyed by digitized battery systems, they can inflict almost arbitrary economic losses on competing spinning generation.
When I first touched on this subject in early 2019, I believed that their freedom to do so could be constrained by the inefficiencies of finite time resolution electricity markets (power is typically sold in chunks of 30 minutes to days, some time in advance), or even regulatory capture by expensive and politically favored gas peaker plants. This turns out not to be the case. Batteries have such perfect control of their net power flow they can “make the market” and conform to, and compensate for, any existing or attempted market imperfections.
As utility after utility, gas plant operator after operator is discovering, batteries seem to operate under different physical laws. Like Neo in the Matrix, they can secretly turn off gravity or dodge bullets if they want to. Even if there’s no apparent reason why they should.
In fact, it’s now clear that the best way to minimize the relative advantage of batteries relative to legacy operators is to move immediately to a real time, automated electricity market, or at least make one available. Regulation is still necessary, in particular to avoid Enron-like shenanigans, but at least with rule transparency gas operators will be able to more easily predict just how much money they’ll be forced to fork over to the battery operators in any given time period. That’s right, their operation is at the pleasure of the battery operators who control market price, and they will have to pay for market access.
Large (>100 MWh) battery systems coupled to synchroverters are arbitrarily more responsive to market fluctuations over shorter timescales than gas generation, pumped hydro, or any other large scale generation or storage mechanism. Speed is king. Inadequately serviced demand for short-term grid stabilization is a market inefficiency and represents an opportunity for arbitrage.
Hard experience has shown that demand variation with “smart metering” is an uphill battle, as consumers naturally resist being “upgraded” to a system which locks them out of the economic benefits extracted by inter-mediating their access to arbitrary quantities of reliable, cheap electricity.
Analogous to High Frequency Trading?
Mismatch between supply and demand in the grid is signaled via tiny shifts in frequency. Coupled to this, local energy spot price has historically been manipulated by gas peaker plants to maximize revenue generation. For example, in Australia prior to the installation of the Horndale (Tesla) battery, the spot price routinely climbed to the cap of $14000/MWh as gas generators delayed ramping up to maximize revenue.
Installing batteries is similar to high frequency trading (HFT) in that it exploits market inefficiencies to both improve the functioning of the market, provide liquidity, and generate revenue.
It is different to HFT in several important ways, however.
There is an advantage to geographic collocation for battery services adjacent to areas of variable supply and demand, due to the fixed and variable costs of long distance power transmission. It’s not quite as extreme as installing the HFT server directly adjacent to the stock exchange, though!
The fixed frequency of the power grid (60 Hz in the US) does not demand incredibly sophisticated high speed electronics, like that used in financial HFT. In other words, the fixed frequency of the grid means that the first generation of batteries will not be rendered obsolete by a subsequent arms race as seen in financial HFT.
Combined, these two key differences to financial HFT mean that there is a strong first mover advantage in the battery grid stabilization space. I contend that this advantage outweighs the risk of future battery systems delivered at lower cost. In fact, the EROI on battery system is so short, and their revenue so high, that the only practical constraint to their deployment today is, and has been for several years now, manufacturing capacity no matter how fast it ramps.
What is the 10-20 Year Outlook?
The renewable energy industry is growing by about 25% per year, led increasingly by solar photovoltaics (PV). Costs are falling by about 10% per year. The principal brake on solar deployment today is a lack of load shifting capacity, which is why a lot of new build solar is Southern California integrates load shifting batteries.
Each battery system, in addition to generating revenue through market arbitrage, also increases the usage duty cycle and marginal value of present and future solar or wind installations.
Battery costs have fallen to the point where in many places (Lazard 2018) it is cheaper to build new solar+batteries than to operate existing gas and coal power systems. Even if gas costs, which are often driven by transport costs, fall, carbon burning grid assets are well on the way to being stranded.
Uber’s main innovation was devising a way for the company and car owners to monetize the accelerated depreciation of their cars. This was devastating to traditional taxi operators who had to fund the liabilities of fleet cars, medallions, and traditional regulation. Uber offered a way for car owners to extract equity from their car, even as they serviced loans on it! That this was rarely a good deal for the drivers long term, or even short term, is besides the point.
Uber’s business model is just one precedent for a new market entrant finding a way to crater the value of a third party’s assets and involuntarily transfer much of that wealth away from the original owner who, no longer able to compete or derive revenue from their infrastructure, must now pay rents to operate it at all. This sort of financial chicanery borders on morally dubious except that no-one forced governments to subsidize gas plants, chain these “assets” to future taxpayers, or generate extremely expensive electricity while exploiting our traditional acceptance of life-endangering pollution of public spaces, provided that the pollution is mostly invisible gases that cook us relatively slowly.
Batteries operating on a given grid are like Superman playing the casinos in Las Vegas. Their market-making speed enables them to play the ultimatum game with less responsive competitors, directly monetizing the involuntarily induced depreciation of their stranded assets.
Not only is there not much that spinning generators can do about battery operators eating their lunch, and the table they’re sitting at, with every additional tranche of battery capacity added the game shifts further out of their favor. This is the key point. Progressive battery installations actively shift the equilibrium of the market further in favor of additional battery storage.
There is no practical limit to battery installation on a given grid due to competition with traditional operators. Batteries are the only game in town, and their installation will proliferate as long as there are gas and coal plants to pilfer, solar power to load shift into the evening, and falling hardware costs to cover costs. The natural end state of this transition is 5-15x solar capacity overbuild (it’s very cheap) with enough batteries to load shift over a diurnal cycle for levels of consumption that currently look quite high, due to future low prices and induced demand. No gas, no coal, no nuclear. Wind, yes, in windy cloudy areas.
Indeed, we’re within a decade of seeing large scale electrical synthesis of natural gas and other hydrocarbons. Electricity is cheap enough that the previous century’s practices of burning fuel to make it will now be run in reverse. Industrial hydrocarbons, used for chemical synthesis, plastics, and jet fuel, will be carbon neutral.
At present rates of growth, non-solar sources of energy will likely be excluded from the market by 2030. As costs continue to fall, fixed asset payback times for solar and batteries will reduce to the neighborhood of 24 months. Ultimately, the price of electricity for the consumer will fall enough to transform the way our society uses energy, opening up further opportunities for industrial growth.
(Non) Impact of Policy
At this point it doesn’t matter particularly what any given competing operator or regulatory body does. Even the Australian federal government won’t be able to ban batteries or solar on some flimsy pretext. Already Australia has more rooftop solar, per capita, than nearly any other country. This infrastructure is already built, and at this point just needs consolidation.
Indeed, if the Australian government insists on being a global cautionary tale, either by failing to encourage their friends in the coal industry to move onto better technology, or finding new friends, they’re only paying out just enough rope to hang themselves. What would they prefer? Robust, publicly owned solar and battery plants within reach of central control, regulation, and taxation? Or the continuing proliferation of totally unregulated virtual local grids with almost all solar and batteries behind the meter, where consumers bear the marginally higher capital costs upfront and then keep their cheap electricity forever, largely out of reach of the official utility concessions?
If Australia’s coal and gas operators don’t like the big batteries popping up like mushrooms, they’re really going to hate the proliferation of batteries behind the meter. There’s not much that policy can do to regulate “demand,” such as that generated by large private networks of batteries doing something like Tesla’s autobidder but with no real limits on just how predatory they can be. And you might have thought that bitcoin mining had a public moral hazard!
Yet More Details
Traditionally, electricity suppliers on the grid have reacted to demand in a way that reflects the lack of elasticity in the market. Broadly speaking, demand varies slowly throughout the day in a predictable way, and there is always enough latent capacity to keep the grid stabilized. During times of peak demand, a lack of supply may lead to a lack of voltage stability, causing rolling blackouts.
Operators of a battery, however, can take more than grid frequency variations and local energy spot prices into account. In particular, the degree to which the battery degrades with different kinds of usage incentivizes more sophisticated modeling of the participation of other actors within the grid.
For example, a battery power supply contract oriented around grid stability would simply charge or drain the battery in response to frequency variations, smoothing out the peaks and troughs. Existing big batteries operating in frequency control auxiliary services (FCAS) markets have done just that.
In contrast, a strategy oriented toward maximum revenue would allow frequency to fall while spot price increased, dumping limited power to maximize the area under the curve. This is much closer to the actual operating procedures of peaker plants in Australia prior to the Horndale battery breaking their monopoly.
Finally, with multiple actors competing for the same profit opportunity, it is necessary to predict when and by how much other players might enter the market. For gas, this is relatively straightforward since the plants take tens of minutes to spin up and spin down and are decidedly non-stealthy.
The key insight here is that it’s possible to enact a strategy that maximizes action uncertainty on the part of other players. Consider a sudden rise in electricity demand caused by, say, a Victorian coal generator tripping. The battery can manipulate the shape of this rise. If the gas peaker plant starts up, the battery can flood the market, diminishing the price and cutting the peaker’s revenue. If the peaker does not start, the battery can enjoy a high level of revenue for as long as the demand persists. If the peaker wants to lock in a given price level, they will have to pay the battery operator a fair market price for this opportunity, or else take their ball and go home. Given that they have precisely no cards in their hand, it’s turned out to be a bit of a shakedown.
A fixed strategy is learnable and predictable, but there’s no reason why the battery operators couldn’t employ an adaptive or randomized strategy like a skilled poker player, optimized to maximize any desired metric. Battery operators can choose between strategies designed to maximize revenue, profit, competitive advantage, or competitor losses, over any given timescale and with almost perfect deniability. The software required to do this is simple enough that it could be coded up in a weekend by any motivated high school student taking a break from generating trading algorithms on Numerai.
For example, if peaker operators in Australia run to the government and extract an operating subsidy to keep them in business, there’s no way they can stop the battery operators from pocketing every last cent. The more federal regulators insist on attempting to tilt the game toward gas or coal, the more they generate market imbalance, the more profit opportunity they create for battery operators. Like a bug trapped in a spider web, struggling only hastens their demise. Best of all, the more the battery profits, the cheaper and cleaner public electricity will ultimately be.
While a sufficiently expert regulator could insist on a set of public standards on metrics, APIs, and falsifiable strategies, it’s highly unlikely that owners of proprietary battery management systems (BMS) would willingly expose that information. Furthermore, any battery activity behind the meter can’t be regulated in the same way without enormous violations of privacy and the right to use services at will.
We already have companies like Nest tweaking air conditioning power consumption across millions of users to participate in grid services. While it’s not beyond the ability of legislators to produce truly terrible and unenforceable laws I’m reasonably confidant that even the Australian federal government would balk at attempting to regulate the right to turn lights off and on at will.
In short, existing precedent has locked in the grim destiny of gas and coal electricity generation in practically every grid on Earth.
Where Do We Go From Here?
It’s not all bad news. Incumbent operators have decades of experience and preferential access to all kinds of very cheap infrastructure-oriented capital. Their stakeholders will never be unhappy about decreased exposure, increased revenue, better environmental credentials, happier customers, and reduced costs of political involvement. All they have to do is pick up the phone, buy their own bigger, better, cheaper battery plants, and join the fun.