Tesla the Narrative Company

DOGE Eats Its Own

There are companies that you value by looking at their profits and cash flow and putting a multiple on them. And then there are companies that you value by listening to their story. You don’t buy the stock because it earned a lot of money last quarter; you buy it because it is building a glorious future of artificial intelligence and robots and transforming humanity, and you want a piece of that. The most famous, and most successful, practitioner of this is of course Tesla.

For years, the story has been that Tesla isn’t really a car company. It’s an AI company that happens to put its technology on wheels. The cars it sells today are just a down payment on a future of fully autonomous robotaxis, a trillion-dollar opportunity that will make today’s car sales look like a rounding error. This is a very good story! It has helped make Tesla the world’s most valuable automaker, even as its actual car business has been, let’s say, a little bumpy.

A good story needs a good storyteller, and it helps if the world’s most powerful people are, if not actively helping, at least not actively trying to tear the story apart. For Elon Musk, this has become a problem. Here is Reuters on his recent falling-out with his one-time ally, President Donald Trump:

"He's upset that he's losing his EV mandate and … he's very upset about things but he can lose a lot more than that," Trump told reporters at the White House on Tuesday.

Though Musk has often said government subsidies should be eliminated, Tesla has historically benefited from billions of dollars in tax credits and other policy benefits...

Trump on Truth Social on Tuesday suggested Musk might receive more subsidies "than any human being in history, by far," adding: "No more Rocket launches, Satellites, or Electric Car Production, and our Country would save a FORTUNE."

Trump later doubled down, telling reporters with a smile, "DOGE is the monster that might have to go back and eat Elon."

This is, for Tesla’s narrative, less than ideal. The feud is remarkable for how efficiently it threatens every single chapter of the company's story, all at once.

First, there’s the chapter about today. Trump is threatening to kill the EV subsidies that have helped Tesla’s current business. This comes at a time when the core car-making operation is already struggling. There is a brutal price war raging in China, started by rivals like BYD, and Tesla’s sales have plunged in key European markets. For years, the company has leaned on the sale of regulatory credits—nearly $11 billion worth—to prop up its profitability. A hostile president threatening your main source of demand is a bad look when the fundamentals are already shaky.

Then there is the chapter about the future, which is where the real money is supposed to be. Tesla’s valuation is built on the promise of robotaxis. But robotaxis require regulatory approval. A president who is publicly feuding with your CEO and controls the US Transportation Department is a significant hurdle. This is particularly true when your robotaxi launch in Austin involves cars driving on the wrong side of the road and has already attracted the attention of federal safety regulators. The path to an autonomous future was already filled with technical and legal potholes; now it’s also being targeted by the White House.

Finally, there’s the chapter about the storyteller himself. The feud, of course, was ignited by Musk’s own political activism. He has become a central character in his company’s drama in a way that is not always helpful. Former employees have started saying the quiet part out loud ("The problem is Elon"), and customers in France are suing because they feel their cars have become "far-right totems." When the author of your grand narrative becomes its biggest liability, the story starts to get very confusing.

The ultimate irony is that a company built on a story of hyper-advanced, futuristic AI finds its fate hanging on something ancient and deeply human: a public spat between two powerful men. There is no line item on a balance sheet for the risk that the president might weaponize DOGE to "eat" your CEO. Presumably this is now a required disclosure in the 10-K.

The Kilowatt-Hour War

The race to dominate AI is, we are told, about who has the smartest models, the best chips, and the most PhDs. And sure, that’s all true. But it might also be about who has the cheapest electricity. The New York Times highlights the two very different energy bets being placed by the world’s two largest economies:

In China, more wind turbines and solar panels were installed last year than in the rest of the world combined. And China’s clean energy boom is going global. Chinese companies are building electric vehicle and battery factories in Brazil, Thailand, Morocco, Hungary and beyond.

At the same time, in the United States, President Trump is pressing Japan and South Korea to invest “trillions of dollars” in a project to ship natural gas to Asia. And General Motors just killed plans to make electric motors at a factory near Buffalo, N.Y., and instead will put $888 million into building V-8 gasoline engines there.

And this isn't some high-minded climate debate. This is about who gets to power the future. AI is an insatiably hungry beast, and its primary food is electricity. The International Energy Agency projects data centers will consume more electricity by 2030 than the entire country of Japan does today. Nvidia figures a trillion dollars will be spent just upgrading data centers for AI in the next few years. All those fancy Blackwell GPUs need to be plugged into something.

This is where it gets interesting. The US strategy is to lean into what it has: lots of oil and gas. It’s an “energy dominance” play built on a fossil fuel infrastructure that has powered the world for a century. China, on the other hand, is making a massive, state-funded bet on dominating the next century’s energy supply chain: solar, wind, and batteries.

If this were just about volume, you could argue the US has a decent shot. But it’s also about cost. And on cost, it’s not even close. In March 2024, the average household electricity price in the US was $0.18 per kilowatt-hour. In China, it was $0.08. Your power bill is more than double theirs.

This matters. A lot. That cost differential is a massive structural advantage for China’s AI ambitions. If you are running an AI factory, your single biggest operating expense, by far, is energy. A lower electricity cost means you can train more models, run more inferences, and undercut your American competitors on price. Chinese firms like Huawei and DeepSeek can afford to be less efficient at the chip level if they are wildly more efficient at the plug level.

And the scale of China’s build-out is hard to comprehend. The US installed a respectable 32 gigawatts of new solar capacity in 2024. But in 2023, China installed more clean energy than the rest of the world combined. If you think that’s wild, Peter Diamandis, chairman of XPRIZE Foundation, projects what’s coming next:

By 2030, China could have the ability to produce enough solar and storage infrastructure each year to match the entire electricity generation capacity of the United States.

Right. China's annual addition could power the entire US. This isn't a gradual shift; it’s a land grab for the future of power generation.

The irony, of course, is that the US is home to the companies designing the AI revolution—the Nvidias and Googles and OpenAIs. But China is building the next-gen infrastructure to run it. For decades, the US outsourced manufacturing to find cheaper labor. Will it risk a future where it has to, in effect, outsource its compute because it can’t afford its own electricity? The great AI race might not be won by the country with the best chips, but by the one with the cheapest kilowatt-hours.

The Pragmatists Win at Apple

It seems there has been a palace coup at Apple. For years, the company has insisted it would solve its AI problems its own way, with its own technology. But Siri is still, you know, Siri. And now, the pragmatists appear to have seized control. The latest sign is a report that Apple is seriously considering swapping out Siri's homegrown brain for an AI model from competitors Anthropic or OpenAI.

This is a story about corporate power dynamics. On one side, you had the AI chief, John Giannandrea, a purist hired from Google to build Apple's AI capabilities from the ground up. He’s the one who, according to colleagues, was skeptical of chatbots and conservative about the pace of AI. On the other side are the product people, like software chief Craig Federighi and new Siri boss Mike Rockwell, who have to ship products that customers don't laugh at. After a year of embarrassing delays and a tepid rollout for "Apple Intelligence," it seems the product people have won. Mark Gurman of Bloomberg reports on the changing of the guard:

The project to evaluate external models was started by Siri chief Mike Rockwell and software engineering head Craig Federighi. They were given oversight of Siri after the duties were removed from the command of John Giannandrea, the company’s AI chief. He was sidelined in the wake of a tepid response to Apple Intelligence and Siri feature delays. …

Still, Federighi, Rockwell and other executives have grown increasingly open to the idea that embracing outside technology is the key to a near-term turnaround. They don’t see the need for Apple to rely on its own models — which they currently consider inferior — when it can partner with third parties instead, according to the people.

This is a fundamental battle for the soul of the company. Apple’s entire mythology is built on vertical integration—it makes the hardware, the software, and the custom silicon to create a perfect, seamless product. To suddenly decide to license the core engine for its next generation of computing from a rival is… well, it’s what Samsung does. Samsung happily slaps "Galaxy AI" on its phones, knowing full well that much of the heavy lifting is being done by Google's Gemini. For Apple to adopt this strategy is a profound admission of defeat.

But it’s a pragmatic one. The internal models are inferior. The talent is leaving for competitors like Meta who are offering telephone-number-sized compensation packages. And the big Siri overhaul is already delayed until at least 2026. If your choice is between shipping an embarrassing product you built yourself or an excellent one you licensed from someone else, the product guys will choose the latter every time.

Of course, Apple being Apple, they are trying to have it both ways. The plan isn't just to use ChatGPT; it's to use ChatGPT running on Apple's own "Private Cloud Compute" servers. They want to rent the brain but own the skull. It's a complex, probably very expensive compromise that allows them to maintain the illusion of control while outsourcing the hardest part. For a company that built an empire on making everything itself, it’s quite a turn. But if you want to keep selling three-trillion-dollars-worth of phones, maybe the soul of your voice assistant is a fine price to pay.

The Scoreboard

• AI: Us Senate Overwhelmingly Rejects Plan to Stop States Regulating AI (FT)
• Semiconductor: Exclusive: Intel’s New CEO Explores Big Shift in Chip Manufacturing Business (Reuters)
• Acquisition: Grammarly to Acquire Email Startup Superhuman in AI Platform Push (Reuters)
• Social Media: X Will Deploy AI to Write Community Notes, Expand Fact-Checking (Bloomberg)