The email arrived on a Tuesday, the way most bad news does. A colleague in Memphis forwarded me a press release — Tesla had deployed the largest grid battery in the United States at xAI's Memphis supercluster — and asked, only half-joking, whether this was finally the decentralized energy future I had been preaching about since 2017.
I read it three times. Then I closed the laptop and sat with it for a while.
By the headline, the news is impressive. By the mechanism, it is a confession. A battery was built not to democratize energy but to shelter one company's artificial-intelligence ambitions from a grid that could not move fast enough to serve them. The neighborhood next door — a historically Black community called Boxtown that has absorbed industrial pollution for generations — was not consulted. Nobody voted. Nobody governed. A battery simply appeared, and the word 'sustainable' was attached to it like a ribbon on a machine that runs on gas.
Build for humans, not just nodes. I have repeated that line in workshops from Prague to Berlin, from Lisbon to a repurposed warehouse where 150 confused developers once asked me whether decentralization was a technology or a religion. Reading the Memphis release, I realized the line cuts both ways. Tesla built a node. The humans were an afterthought.
I want to be honest about what follows. The original report this piece responds to was almost comically thin. Five information points, three of which were the article's own opinions about 'sustainable energy transitions' and 'regional energy dynamics.' Exactly one hard fact: Tesla built a large battery at an xAI site. No megawatts. No megawatt-hours. No chemistry. No cost. No policy context. No timestamp. That sparseness is itself the story, and I will treat it as such — an analytical framework for what this event should have been measured against, and a warning about how the clean-energy narrative gets used in the middle of a bull market.
The Memphis site sits at the intersection of three of the most concentrated forces in the modern economy: artificial intelligence, energy, and capital. It is also, unavoidably, a story about decentralization — the idea that brought me into this industry and the idea this event quietly betrays. So let me do what a protocol engineer does when handed a marketing claim: take it apart, look at the wiring, and ask who actually holds the keys.
The first thing to understand is what this battery physically is, because the word 'battery' hides an entire industry of design decisions.
There are two battery universes that share a name and almost nothing else. The first is the automotive one: high energy density, fast charging, nickel, cobalt, expensive, engineered to move two tons of metal down a highway. The second is the stationary one: indifferent to weight, obsessed with cycle life and cost-per-kilowatt-hour, engineered to sit still for twenty years and absorb punishment. These are not the same product, and confusing them is the single most common error in coverage of events like Memphis.
For grid-scale stationary storage, the industry has converged almost entirely on lithium iron phosphate — LFP. The reasons are mechanical, not ideological. LFP cycles roughly six to eight thousand times before degrading to eighty percent of its original capacity. A nickel-manganese-cobalt cell manages perhaps three thousand. LFP contains no cobalt and no nickel, which means it sidesteps the geopolitical traps of the Congo and Indonesia. And because lithium prices collapsed — from around six hundred thousand yuan per ton in 2022 to roughly seventy to ninety thousand by 2024 and 2025 — the raw-material floor for LFP systems kept falling just as demand exploded.
Tesla's Megapack, the product almost certainly at the heart of this event, is a containerized LFP system. Each unit holds roughly 3.9 megawatt-hours, is liquid-cooled, and rolls off lines in Lathrop, California and Shanghai. When a press release says 'the largest battery,' and the vendor's flagship product is a Megapack, you are reading about an LFP array — not some exotic new chemistry, and certainly not the powertrain tech that made the company famous.
Which raises the question the original report never asked. Where exactly is this battery connected?
Here is where the coverage became misleading, and here is where I want you to slow down with me.
The phrase 'grid battery' conjures an image of a utility-scale power plant feeding the public grid, absorbing wind and solar at noon and discharging it at dusk. That is one architecture. There is another, and it is far more likely here: behind-the-meter storage.
Behind-the-meter means the battery sits on the customer's side of the meter. It serves the customer first. It smooths the data center's load, shaves its peaks, provides backup when the connection flickers, and buffers the moment when a cluster of GPUs spikes demand faster than the grid can respond. This is not a public good. It is private infrastructure dressed in public language, and the difference is not semantic — it determines which tax credits apply, which regulators have jurisdiction, and who bears the cost when something goes wrong.
The distinction matters because the economics are completely different. A standalone grid asset earns money by arbitraging wholesale power prices and selling ancillary services into organized markets. A behind-the-meter asset earns money by lowering one company's bill and de-risking one company's operations. The first is a business. The second is insurance. The press release blurred them into a single word, 'grid,' and let the reader imagine the more generous of the two.
Education is the ultimate yield, I tell people at every workshop, and this is exactly the kind of distinction that education is for. When a headline says 'grid battery,' ask: whose grid, whose meter, whose risk? The answer to those three questions tells you whether you are reading about infrastructure or about a very expensive uninterruptible power supply with good public relations.
Now the claim that should have been audited before it was published: 'largest in the United States.'
I have spent enough time around deployment data to be allergic to superlatives without a denominator. So let me supply the denominator the original piece omitted. The largest grid battery in the United States has, for years, been Vistra's Moss Landing facility in California — roughly 750 megawatts and 3,000 megawatt-hours, and yes, it has a fire history that regulators are still working through. Close behind sits NV Energy's Gemini project in Nevada, at roughly 380 megawatts and 1,400 megawatt-hours. These are public, sited, interconnection-approved utility assets.
The Memphis installation, by every public account I can find, is in the range of a hundred-plus megawatts. That is a serious engineering achievement. It is not the largest in the country. The superlative almost certainly means one of three much narrower things: the largest Tesla-built site, the largest behind-the-meter storage in the country, or the largest battery installed at a single private facility on the fastest timeline. Each of those is a real record. None of them is 'largest in the United States,' full stop.
You cannot tokenize your way out of a physics problem. You also cannot footnote your way out of a superlative. When a statistic arrives without a scope — largest what, measured how, compared to whom — the honest move is to halve it and wait for the correction. That is the discipline I try to teach, and it is the discipline the original coverage skipped entirely.
But here is where the deeper story begins, and it is not about capacity. It is about time.
The reason a company would build a gas-and-battery microgrid instead of simply plugging into the public grid is that the public grid cannot say yes fast enough. This is the structural fact that the entire 'sustainable energy' framing conceals.
In the United States, the interconnection queue — the line of projects waiting to connect to transmission — has been catastrophically backed up for a decade. Lawrence Berkeley National Laboratory's queue studies have shown median waits stretching into multiple years, with some regions approaching five. To connect a load the size of a large AI data center, you do not join a queue. You join a pilgrimage. You hire lawyers. You negotiate with a regional transmission organization, a utility, and often several municipalities whose consent is neither guaranteed nor quick.
The workaround the industry invented is called 'bring your own power.' You stop waiting for the grid and you build generation yourself. You roll in mobile gas turbines, you sign a supply agreement with the local utility for whatever buffer it can spare, and you install a large battery to make the whole thing behave. The battery is not there to decarbonize anything. The battery is there to turn a chaotic, spiky GPU load into something the turbines and the grid connection can tolerate without tripping.
That is the real function. The battery is a shock absorber. It is a device that buys time — literally, in milliseconds, and figuratively, in years of avoided interconnection. Its economic value is the value of schedule compression. When you are racing competitors to train the next model, a battery that lets you power on eight months earlier is worth vastly more than the electricity it stores.
So when I read that this event represents a 'transition to sustainable energy,' I want to ask the only question that matters: transition from what?
The answer, visible to anyone who reads the local reporting, is natural gas. The Memphis xAI site is powered substantially by mobile gas turbines. The battery rides alongside them. Environmental organizations in the region, along with the NAACP and the Southern Environmental Law Center, have raised concerns about the turbine emissions and the way the facility was sited relative to the Boxtown community. Those turbines produce electricity at a carbon intensity of roughly four hundred to five hundred grams of CO2 per kilowatt-hour, an order of magnitude higher than solar or wind over their lifecycles.
I want to be fair here. Batteries genuinely do reduce certain emissions — they let you run turbines at more efficient steady states, they capture otherwise-curtailed renewable energy, and they displace dirtier peaker plants in some markets. That is real. But the phrase 'sustainable energy transition' applied to a facility whose baseload is fossil generation is not a technical description. It is branding, and it works precisely because the audience is not equipped to separate the battery from the system it serves.
This is the moment where my blockchain background stops being a tangent and becomes the actual lens. Because the story I am describing — centralized infrastructure, marketed as a collective good, governed by a handful of insiders — is the exact failure mode the crypto movement was founded to correct. And watching it repeat in energy should terrify anyone who cares about decentralization.
Let me be specific about the parallel, because I do not want to traffic in vibes.
For years, a cohort of builders — Energy Web, Power Ledger, Brooklyn Microgrid, and a dozen DePIN projects — have argued that energy's future is decentralized: rooftop solar, home batteries, electric vehicles, and smart inverters, all coordinated by a distributed ledger instead of a utility command center. The pitch was elegant. Distributed energy resources are proliferating anyway, so you need a coordination layer, and a public chain is the most neutral coordination layer available. Households become prosumers. Communities govern their own microgrids. Value flows to the edge.
Then you look at events like Memphis, and you realize the market went the other direction. The largest battery in America was not built by a community. It was built by the richest man in the world, behind his own meter, to serve his own compute. The decentralization that actually scaled was not the blockchain kind. It was 'a single firm vertically integrating around a bottleneck.'
I do not say that to gloat. I ran the experiments that failed. When I led the Aave whitepaper translation project in 2020, demystifying liquidation mechanisms for five thousand non-technical users, I believed that comprehension and access would drive adoption. When I curated the 'Art & Algorithm' gallery in Prague in 2021, I believed provenance would outcompete speculation. Some of those bets worked. The energy ones, by and large, did not, and the reason is instructive: energy's binding constraint is physical and regulatory, not informational. No token clears an interconnection queue. No DAO installs a transformer.
This is the hardest lesson for my industry, and the one I keep circling back to. Decentralization is not a property you can bolt onto a system by adding a ledger. It is a practice, and it requires the boring, unglamorous work of governance, ownership, and consent — the exact things the Memphis facility skipped.
Consider Autobidder. Tesla's energy-trading software is the quiet crown jewel of its energy business. It takes control of a fleet of batteries and bids them into wholesale electricity markets, buying low, selling high, stacking ancillary services, optimizing across milliseconds. It is, functionally, an algorithmic market participant. And it is proprietary, closed, and owned by one company.
Think about that for a second. The promise of decentralized energy markets was that no single actor would control pricing. The reality is that the most sophisticated participation in those markets runs through a black-box optimization engine owned by a vertically integrated manufacturer that also makes the batteries, the software, and increasingly the generation. Vertical integration wins on velocity. It also re-concentrates the exact power the decentralization movement claimed to disperse.
I saw a version of this in the DAO space. On-chain governance turnout sits perpetually below five percent, while a handful of whales and venture funds steer proposals. The architecture is decentralized. The control is not. The Memphis battery is the energy equivalent: a modern, capable machine, governed by a committee of one.
And then there is the carbon accounting, which is where the branding does its most effective work.
When a facility reports its climate impact, the numbers are structured into scopes. Scope 1 is direct emissions — the gas turbines. Scope 2 is purchased electricity. Scope 3 is everything upstream and downstream. A battery does not sit in any of these scopes as a negative emission. It is equipment. Its climate benefit, if any, shows up only as avoided emissions elsewhere, which is a modeling choice, not a measurement.
So when a press release invites you to read a battery as a 'transition,' it is quietly asking you to skip Scope 1 entirely. The turbines keep running. The megawatt-hours keep coming out of fossil fuel. The battery redistributes when that electricity is consumed and how efficiently, but it does not change that it is consumed or what generates it. Reporting the battery while omitting the turbines is not a rounding error. It is the whole ledger, minus the debit side.
xAI is a private company. That matters more than it sounds. Private firms have weaker disclosure obligations than public ones, and their environmental reporting is largely voluntary. So the one sentence in the original coverage that was an opinion — that this represents a sustainable transition — is also the one sentence that cannot be verified with public data, because the operator is not required to publish the data that would settle it. We are asked to trust the narrative precisely where transparency is thinnest.
Now let me flip the lens, because I do not want to write a screed. The contrarian reading of Memphis is that the decentralization advocates got the future wrong, and we should say so out loud.
For a decade, the crypto-energy narrative assumed that the path to a cleaner, fairer grid ran through consumer markets and token incentives. Give households a reason to install batteries. Reward them for discharging at the right time. Let microgrids trade. That story was intellectually satisfying and commercially marginal. What actually happened is that the demand shock for energy came from the top — from AI — and top-down capital answered it in months, not decades.
The brutal lesson is that decentralization is a luxury of abundance. When a resource is plentiful, you can afford to distribute it slowly and govern it collectively. When a resource becomes a bottleneck — and electricity, transmission capacity, and transformers are all now bottlenecks — power concentrates around whoever can move fastest. AI turned electricity into a bottleneck. Bottlenecks produce monopolies.
This is where the blockchain crowd should pay attention rather than moralize. The same dynamic is playing out in compute. When I advised the EU regulatory task force on decentralized governance in 2025, drafting 'Community First' standards for dispute resolution inside smart contracts, the most contested clause was not about code. It was about who has standing when a system is privately controlled but publicly consequential. That is exactly the Boxtown question. A facility appears. It changes the air. It changes the grid. It changes who pays for transmission upgrades. And the community with standing is the community that was never asked.
We built governance tooling for tokens and forgot to build it for the physical world.
The optimistic, and I think correct, framing is this: the AI-driven energy boom is going to force the governance question to the surface, whether or not the crypto industry is ready to answer it. Every campus like Memphis raises the same issues — local consent, cost allocation, emissions disclosure, and grid fairness. Those issues are solvable, but only with institutions, not with whitepapers. The next decade of decentralization will not be won by whoever ships the cleverest token. It will be won by whoever can credibly answer the question, 'who decided, and were the affected people in the room?'
I ran a bear-market peer-support network in Prague in 2022 called Reclaim, for two hundred burned-out developers. What I learned there is that resilience is not a matter of optimism or pessimism. It is a matter of whether people have real agency over the systems their lives depend on. By that standard, a battery behind a fence is not resilience. It is dependency with nicer optics.
So what should a careful reader take from a single, badly-sourced press release about a battery in Tennessee?
Take the method, not the number. When you see a superlative, demand a scope. When you see 'grid,' ask whose meter. When you see 'sustainable,' ask where the gas turbines are. When you see a record tied to a private company in a bull market, assume the framing has been optimized for investors and adjust accordingly. None of that requires pessimism. It requires the same skepticism you apply to a smart contract audit: trust the parts you can verify, and price the parts you cannot.
And take the warning. The story of Memphis is the story of this cycle. In a bull market, capital stops asking whether a thing is decentralized and starts asking only whether it is fast. The fastest path to energy for the most impatient industry ran straight through fossil fuel, private meters, and an unconsulted neighborhood. Decentralization, as a marketing word, survived. As a practice, it was nowhere in the room.
That should not make us cynics. It should make us better builders. Build for humans, not just nodes. Education is the ultimate yield. And when someone hands you a battery and calls it a movement, ask the only question that has ever mattered: who gets to decide, and who was never asked?
The next grid will be built in a hurry. The question is whether it will be built by a committee of one, or by the people who have to live beside it. On the answer to that, everything — the emissions, the economics, and the meaning of the word decentralized — still depends.
