Here’s what’s on our plate today:

• 🧪 How AI's electricity boom is turning EVs into grid assets.

• 📰 NASA picks Schmidt's rocket company, Snap bets on fashion smart glasses, and China's festival shows AI and weak demand.

• 🛠️ Weekend To-Do: check your EV's V2G support, explore a virtual power plant, map your energy exposure.

• 🗳️ Poll: Will automakers' EV-as-grid-asset bet work?

Let’s dive in. No floaties needed…

The Laboratory

TL;DR

• Power is the new bottleneck: AI data center electricity use grew 50% in 2025, and Microsoft has committed roughly $80B to AI data center buildout while holding purchased hardware it can't power.

• GM's pivot runs on stranded assets: After $7.1B in EV charges, GM adopted the vehicle-to-grid standard across 250k bidirectional vehicles and partnered with Peak Energy on sodium-ion grid storage.

• The contribution is modest: Every GM EV discharging at once is roughly equivalent to one mid-sized data center. The real value is smoothing demand peaks, not powering AI directly.

• Participation is the gamble: Owners trade 9 to 14% added battery degradation for wholesale-rate payments, and 55% already cite lost flexibility as a barrier.

• The stakes: The people paying AI's electricity costs and absorbing its job disruption are the ones being asked to power it.

How AI's electricity boom is turning EVs into grid assets

For much of the past few years, discussions about artificial intelligence have revolved around the idea of abundance. Advocates have argued that increasingly capable AI systems could unlock unprecedented gains in productivity, accelerate scientific discovery, and help create a future in which intelligence itself becomes abundant.

Yet while the industry was busy imagining that future, it was running into a far more immediate constraint: the physical world.

Over the past three years, the world's largest technology companies have spent hundreds of billions of dollars building artificial intelligence infrastructure. They have bought chips, leased land, expanded cloud capacity, and financed enormous data center projects. But despite all that spending, many are running into a more basic limitation: electricity.

The facilities that train and run AI models consume vast amounts of power, and demand is growing faster than utilities can expand generation and transmission capacity. Microsoft has committed roughly $80B to AI data center buildout, while holding purchased hardware it can't power. For the first time since the dot-com era, the bottleneck on a major technology boom is not capital, talent, or silicon; it is the grid itself.

When a shortage becomes that severe, it begins to reshape industries far beyond the one that created it. Assets designed for one purpose suddenly acquire value in another, and this realization is now pushing companies to reevaluate technologies once viewed as niche energy experiments as potential infrastructure.

This is how General Motors, a company better known for selling pickup trucks than powering data centers, ended up arguing that the future of artificial intelligence may depend, in part, on the cars parked in people's driveways.

The company's pitch rests on a simple premise: if power generation cannot expand fast enough to meet demand, the next best option is to make better use of existing energy storage.

The grid becomes the customer

What GM announced at its 'Empower 2026' event was a three-part energy strategy built almost entirely from assets it already owned. The company made vehicle-to-grid capability, the ability of an EV to send stored electricity back into the grid, a standard feature across its EV portfolio, covering the more than 250k bidirectional-capable GM vehicles already on U.S. roads, with pilots underway alongside Pacific Gas & Electric in Northern California and DTE Energy in Michigan. It announced a partnership with Peak Energy, a Colorado battery startup, to develop sodium-ion cells purpose-built for grid-scale storage, with commercialization targeted for after 2028. And it expanded a program with recycler Redwood Materials to repurpose degraded EV battery packs as stationary storage at its own facilities.

Viewed in isolation, the announcements appear to be a conventional energy diversification strategy. In reality, they are inseparable from the challenges facing GM's electric vehicle business itself.

The company took $7.1B in charges in late 2025 as it dismantled the electric expansion it had spent three years building, after the $7,500 federal purchase credit expired and U.S. EV sales fell 4% in 2025, according to CNBC. That contraction left GM holding battery factories, an energy software platform, and a 250k bidirectional vehicles whose original business case had weakened. The energy strategy is what happens when a company looks at those same assets through the lens of the grid's new scarcity rather than the showroom's old demand. The cars GM could not sell fast enough as transportation are being recast as infrastructure.

Turning parked vehicles into power assets

However, that shift in positioning only works if EVs can provide something the grid genuinely needs. Therefore, understanding the appeal of GM's idea requires understanding what vehicle-to-grid technology actually does.

An EV battery is a large electrochemical storage unit, and in a standard charging arrangement, electricity flows one way, from the grid into the car. Bidirectional charging reverses the flow: when a compatible vehicle is plugged into a compatible charger, its battery management system communicates with the grid, and when demand spikes, the system can command the car to export stored power back into the local network. GM's implementation runs through its PowerShift charger hardware and an app-based platform that manages energy flows between the vehicle, a home battery (if present), and, eventually, the grid itself. The concept, often called a 'virtual power plant' when aggregated across thousands of vehicles, has existed in energy research for roughly two decades. It never arrived at scale because nothing forced the issue. The surge in AI demand is the forcing event.

The scale problem

Although the concept sounds transformative in theory, the challenge is that AI's electricity demand is so large that even hundreds of thousands of vehicles can make only a modest contribution.

A single large data center draws anywhere from 100 to 500 megawatts continuously, while the average EV battery holds 60 to 100 kilowatt-hours of usable energy. Even if every bidirectional GM vehicle in the country discharged simultaneously, the combined output would roughly match that of a mid-sized data center. Which means that the honest version of GM's pitch is indirect: EVs cannot power AI, but they can smooth the peaks in regional demand that AI sharpens, freeing utilities to dedicate firm generation to data centers that need constant load. The sodium-ion partnership with Peak Energy, however, is the more direct play. Those cells, built from one of the most abundant elements on Earth, sacrifice the energy density a car needs for cycle life and the low cost a stationary installation requires, and GM is explicitly targeting data centers and utilities as customers, per CNBC.

Even if the technology proves valuable to utilities, its success ultimately depends on participation. A virtual power plant is only as large as the number of vehicle owners willing to join it.

For the consumer, the meaning of the technology is a trade. Participating owners earn payments from utilities for exported power, typically at rates closer to wholesale than retail. In exchange, they accept added wear: peer-reviewed research in Applied Energy found V2G increases battery degradation by 9 to 14% over 10 years. The math can work for a light driver in a high-rate market and fail for a commuter. Drivers appear to understand this instinctively: a 2025 study in Energy Policy found that 55% of surveyed EV users cited the loss of flexibility as their main barrier to participation, ahead of battery degradation and data privacy. The fleet only becomes infrastructure if its owners volunteer it.

A model that other automakers may follow

The question is significant because GM is unlikely to be the last automaker to pursue this model. The forces that produced its strategy are affecting the industry as a whole.

The same stranded-asset logic that GM applied also applies across Detroit and beyond, and the moves are already visible. Ford is spending $2B to convert a Kentucky battery plant originally built for EVs into an energy storage products facility, according to CNBC, while Tesla's energy division, a decade old, now carries margins roughly double its automotive business. Every automaker that overbuilt battery capacity for an EV transition that slowed is staring at the same arithmetic: the grid is the customer that showed up when the car buyer did not. If GM's V2G program demonstrates that a consumer fleet can be monetized as grid capacity, the model becomes an industry template rather than a company strategy.

Taken together, these efforts point toward a broader redefinition of what an EV battery is. For most of the past decade, automakers treated batteries primarily as automotive components. Increasingly, they are viewed as energy assets mounted in vehicles.

The people holding the plug

This brings the story to the people holding the plug. The IEA notes that communities are increasingly pushing back against data center projects, with concerns about affordability and the broader effects of AI on jobs rising alongside them. The result is that several of the most consequential debates surrounding AI are converging on the same group of people.

The person whose electricity bill has been inflated by data center demand, whose town may be contesting a data center proposal, and whose work is being reshaped by the models running inside those buildings is the same person now being asked to buy the vehicle, install the charger, and lease the battery in the driveway back to the grid that keeps it all running. The protest sign and the power cord end up in the same hands.

Whether enough of those hands take the deal will decide if GM's pivot is a structural transformation or a well-framed repositioning of assets it could not otherwise justify. Regardless of how such a technology is adopted, what remains open is whether the people who own the cars, already paying for AI's rise in their bills and absorbing it in their working lives, will agree to power it, too.

Friday Poll

Weekend To-Do

• Check if your EV can do V2G: Look up whether your vehicle and charger support bidirectional charging on your automaker's site; the GM Energy page is a good model for what to look for.

• Explore a virtual power plant: See if your utility runs a VPP program. Tesla's Virtual Power Plant is the clearest consumer-facing example of how the trade works.

• Map your own energy exposure: Pull your last year of electricity bills and check the rate trend; you'll see firsthand why data center demand is becoming everyone's problem.

Headlines You Actually Need

• NASA picks Schmidt's rocket company: NASA selected Eric Schmidt's rocket venture for a Mars mission, setting up a direct race with SpaceX for the future of interplanetary travel.

• Snap bets on fashion smart glasses: Snap is repositioning its Specs as fashion-forward wearables, betting that style, not just tech, is what finally makes smart glasses mainstream.

• China's festival shows AI, weak demand: China's mid-year shopping festival highlighted both soft consumer demand and AI's rising role in commerce, a mixed signal for the world's second-largest economy.

Meme Of The Day

The Toolkit

• Regie.ai: AI sales agent that researches prospects, writes personalized outbound at scale, and handles follow-ups so reps can focus on closing.

• Replit AI: A browser-based coding environment with an AI agent that builds, debugs, and deploys apps from prompts, with no local setup required.

• Sourcegraph: Code intelligence platform with Cody, an AI assistant that understands your entire codebase so it can answer questions and write code that actually fits.

Rate This Edition