Tesla’s record quarter comes with a cost

Revenue surged on strong demand before U.S. EV tax credits expired, yet profit and margins were squeezed

By Shannon Carroll9 min readUpdated October 23, 2025

Tesla’s record quarter comes with a cost

Photo by Ina Fassbender/AFP via Getty Images

Tesla $TSLA’s third-quarter results, released Wednesday afternoon, gave investors plenty of reasons to cheer — at least at first glance. The company beat expectations with $28.1 billion in revenue, up roughly 12% from a year earlier, and, as previously reported, 497,099 vehicle deliveries, the highest in its history. It also generated a record $4 billion in free cash flow and ended the quarter with $41.6 billion in cash and investments — plenty of fuel for its AI and energy ambitions. Those headline numbers suggest momentum: Tesla is still scaling production at a level that most automakers can only envy.

But the bottom line tells a different story.

Net income dropped about 37% year-over-year to $1.4 billion, a sign that the company’s growth is coming at a steepening cost. Tesla’s gross margin came in at around 18%, down from nearly 19.8% a year ago, and far below the 25% range that once defined its peak profitability; operating margin slid five points from a year ago to 5.8%. Sequentially, margins were flat, suggesting that cost pressures have stopped worsening but haven’t started easing, either — a small relief for investors watching for signs of stabilization after six straight quarters of compression.

Adjusted EPS landed at $0.50, a miss versus the Street’s expected $0.55 per share. And operating expenses jumped 50% as Tesla kept spending on software and next-gen manufacturing, a sign that CEO Elon Musk is still flooring the accelerator on AI factories, even as core auto margins thin. Profit pressure wasn’t existential — just persistent — and it’s where the Street will keep grading.

Tesla’s finance chief, Vaibhav Taneja, said capex will “increase substantially in 2026” to fund AI and robotics expansion, underscoring that near-term margin strain is financing longer-term bets. Still, the company declined to give full-year guidance and leaned into “policy uncertainty” language — a tone check that helps explain the stock’s drift: Shares slipped close to 1% in after-hours trading as investors digested a margin picture that looks weaker than expected.

For a stock still trading at a tech-multiple, that silence matters because high-growth valuation and vanishing guidance rarely coexist comfortably.

Margins meet reality

The margin erosion isn’t just arithmetic; it reflects a shifting balance inside Tesla’s business model. The company leaned heavily on pricing incentives to drive volume this quarter, especially as the $7,500 U.S. EV tax credit expired at the end of September. Production totaled roughly 447,450 units, leaving a 50,000-car gap between output and deliveries — the widest since early 2023 and a sign that demand was pulled forward rather than organically growing.

Musk warned on the earnings call that the credit-driven rush may distort the next quarter’s delivery rhythm — “a lot of people moved faster than usual,” he said, adding that production expansion would resume now that he sees “clarity on achieving unsupervised full self-driving” — perhaps a greenlight for another round of factory expansion, reversing months of operational restraint.

Automotive gross margin excluding regulatory credits came in around 15.4%, a shade under expectations and another sign that core profitability is still sliding. Put together, Tesla sold more cars and captured more revenue — but captured less profit per dollar of sales. Layer on rising costs tied to new-model development, robotics, and its energy-storage business, and the math turns more complicated.

Deliveries rose across all regions — up 33% in China and 28% in North America — though executives cautioned that some of that strength was, yes, front-loaded by the expiring tax credits. Tesla also noted lower fixed-cost absorption, higher tariffs, and less benefit from materials deflation — all of which lifted its average cost per vehicle. The company said total tariffs cost roughly $400 million in the quarter — about half of it hitting the automotive business.

And just before earnings, Tesla recalled about 13,000 vehicles (2025 Model 3 & 2026 Model Y) for a battery-pack contactor defect that could cause a loss of propulsion. Concurrently, the National Highway Traffic Safety Administration probe into around 3 million FSD-equipped Teslas (over traffic-signal violations) remains open. While these probes didn't derail Tesla’s results, they keep cost, risk, and brand-overhang in the conversation.

One of Tesla’s quiet profit engines — the sale of regulatory credits to other automakers — also lost power. That revenue line fell roughly 44% year-over-year to $417 million, down slightly from $439 million last quarter, erasing a cushion that has helped Tesla offset pricing pressure in past quarters. Analysts expect that cushion to keep fading as policy shifts narrow the pool of buyers. That’s a small number on the income statement but a big one for margin optics, and with that cushion shrinking, the Street is again asking how Tesla plans to defend profitability without it.

The result is a company that’s growing fast, but with less leverage on every dollar it brings in. It’s an awkward position for Tesla, which has spent the better part of the past decade convincing Wall Street that it isn’t a car company, but a technology platform in motion. That narrative now meets arithmetic: The platform premium only holds if margins recover.

AI, robots, and what comes next

The numbers didn’t break Tesla’s story, but they did shift it — from one of limitless growth to one of cost discipline and long-term conviction.

Tesla has spent years saying that the car business is only the company’s foundation and that the real prize is autonomy, robotics, and software. Musk opened the call by declaring that Tesla had reached “a critical inflection point” as it brings “AI into the real world,” describing the company as having “the highest intelligence density of any AI out there in the car.”

In this latest quarter, Tesla started to move that AI story from the slide deck to the street level.

On the call, Musk described the coming robotaxi rollout as “a shockwave,” saying Tesla is “just at the beginning of scaling quite massively full self-driving and robotaxi.” He expects “no safety drivers in at least large parts of Austin by the end of this year” and plans to expand to “about eight to 10 metro areas” by the end of the year. Tesla CFO Taneja added that the company is already running robotaxi pilots in Austin and the Bay Area, expanding coverage threefold in Austin, with customers logging more than six billion miles using FSD Supervised.

The company also opened a U.S. and Canada waitlist for its robotaxi iOS app, and inked a semiconductor manufacturing deal with Samsung to produce AI-training chips. Musk confirmed that Tesla is designing an AI5 training chip, claiming it will be “40 times better than AI4,” and that both Samsung (Texas) and TSMC $TSM (Arizona) fabs will manufacture it. He said he’s spent “almost every weekend” with the chip design team and called it a “beautiful chip.”

The project also positions Tesla to rely less on Nvidia $NVDA’s supply chain — a vertical-integration play that doubles as an identity statement — and noted Tesla deleted legacy GPU and ISP components to improve efficiency. Tesla also said its internal training cluster, “Cortex,” has scaled to the equivalent of 81,000 H100 GPUs — a concrete marker of how far it’s built its in-house compute stack.

There’s a gap between signaling and scaling. Investors are still pressing for dates, miles driven, take-rate data, cost per robotaxi unit, and profitability timelines before they’ll re-rate Tesla as a software platform rather than a cyclical manufacturer. Until those metrics are less aspirational and more concrete, Tesla’s premium valuation leans on faith. And faith is fine — until the numbers ask for execution.

Musk also teased Tesla’s humanoid robot, Optimus, saying it has the potential to be “the biggest product of all time.” He said version 3 will be ready to show in early 2026 and will look so lifelike that “you’ll need to poke it to believe it’s actually a robot” because it’ll seem like “a person in a robot suit.” The project, he said, borrows heavily from Tesla’s real-world AI work in vehicles — the next logical extension of the platform narrative. Musk also acknowledged the scale of that challenge, noting there’s “no supply chain for humanoid robots,” and that Tesla will have to manufacture “very deep into the supply chain” to make it real.

The CEO kept referencing a “robot army” and tied it to the reason he wants more voting control within the company: “If we build this robot army, do I have at least a strong influence over that robot army?” he said.

Musk also reframed Tesla’s mission in broader terms, calling it “sustainable abundance” — which he defined as a world where clean energy, automation, and AI together eliminate scarcity. “With Optimus and self-driving,” he said, “you can actually create a world where there is no poverty.”

Energy storage: the hidden muscle flex

While the automotive business commands headlines, the energy segment may be the one to watch for what comes next. Musk argued that grid-scale batteries could “effectively double the energy output of the United States without building new power plants,” pitching Tesla’s storage products as critical to national infrastructure.

Energy generation and storage revenue climbed 44% to $3.42 billion on record storage deployments of 12.5 GWh, and Tesla said energy gross profit hit a record $1.1 billion, showing that Tesla’s grid-side ambitions are showing real results. Tesla said demand for its Megapack units is now being fueled partly by the AI boom itself, as data-center operators and utilities look for grid-stabilizing storage. The company expects its new “Megablock” product to begin shipping next year from Houston.

This matters because Tesla’s narrative for 2026 leans heavily on “beyond cars.” If energy continues to scale and deliver high margins, it offers an alternative growth engine when the auto business starts to mature — or face margin compression from competition and price pressure.

Still, Taneja noted that energy faces “headwinds… given the increase in competition and tariffs,” even as margins remain strong — an acknowledgment that its fastest-growing unit isn’t immune from the same pressures squeezing autos.

The long game

Tesla’s third quarter captured the company in split-screen — one frame showing margin compression and operational noise, the other showing a CEO doubling down on scale and ambition.

“You can think of Tesla as, like, a dozen startups in one company,” Musk said, “and I’ve initiated every one of those startups. ... I literally just say, ‘Hey, we’re going to start this thing.’ I post that on Twitter $TWTR, now X, and [say] ‘Join us if you’d like to build it.’”

Whether investors buy what Tesla and Musk are selling remains the question. The story Musk told on Wednesday was utopian; the numbers he reported were not. Then again, Musk’s conviction has never been calibrated quarterly. Each new line of business, from chips to humanoid robots, widens the distance between what Tesla is and what its valuation implies it might become. All of it underscores the same tension running through Tesla’s story: execution measured in quarters, ambition measured in decades.

If the arithmetic catches up to the ambition, this quarter may read, in hindsight, as the hinge between the carmaker Tesla was and the infrastructure company it’s trying to become. Tesla’s ambitions have never been bigger, but neither have its execution risks. The company is trying to scale not just production, but imagination — and the market is still deciding how much of that to price in.

The essential business news, delivered fresh every morning.

Join 500,000+ readers who start their day with Quartz.

By subscribing, you agree to our Terms of Service and Privacy Policy.