Tesla Cut Its Spending and Beat Expectations on Free Cash Flow, While Meta Committed to Even Higher Spending

The panel opened with Tesla's quarterly report, which beat earnings expectations. The bigger story, the hosts said, was spending. Tesla reported free cash flow of about $1.5 billion, compared with a burn of $4.9 billion the prior year. Capital expenditure came in around $4.7 billion, against roughly $7.9 billion that analysts had expected.

Tesla also cut its full-year capex guidance to around $20 billion from $25 billion. The CFO cited using deployment capital more efficiently and funding some of the spending from operating cash flow. One host said it was a head scratcher, because many people expected spending to keep rising.

A panelist said there is concern that, while lower capex is good for profitability, Tesla may be missing out on the AI spend. The projects named were Optimus, the humanoid robot; robotaxis; and Terafab. The panelist asked whether the company should be putting money into those projects.

The discussion treated this as the main concern behind the cut: the gain in profitability is acknowledged, but the question of AI investment is seen as the larger issue.

One panelist recalled analyst Dan Ives saying that Optimus robots would be in everyone's homes within two to three years. The panelist said they replied that they did not want one in their house, and questioned whether the robots would be in anyone's home at all, while noting that some people in Silicon Valley would embrace the idea.

The panelists said the prediction implicitly assumes Musk will meet his aggressive timelines. They pointed to Musk's stated goal of producing one million Optimus units a year, roughly 20,000 a week, as a very ambitious manufacturing target, especially given repeated deadline delays. One panelist compared it to SpaceX launch targets, which they called ridiculous.

The panel noted that Musk has said demonstrations from other companies, which circulate on social media, show pre-programmed moves. One panelist said this suggested the robotics sector may be less advanced than it should be.

The panelists cited Rivian CEO RJ Scaringe, who started a separate company called Mind Robotics. Scaringe said the company is focused on industrial robots, not humanoids, because humanoids are too difficult. The panelists said industrial robots that work on factory floors for specific uses are more deployable and more efficient to spend on, and that the complexity of the human hand is still being worked out.

Tesla shares were down that morning. A host noted that, setting capex aside, the quarter looked good: earnings per share beat estimates, and the surprise free cash flow was linked to lower capex. The automotive gross margin also beat estimates, both with and without regulatory credits.

One area came in well below estimates: energy generation and storage revenue, which had long been a strength for Tesla. The host said the era of strength seems to have ended. The host added that this was not a huge disappointment, since it did not seem to be what the stock was pinned to.

Morningstar analyst Seth Goldstein said the stock is weighed down not just by battery growth but also by auto deliveries. He said Tesla reports its longest backlog ever and cannot make enough cars, with FSD driving demand. Other EV markets are doing quite well. Battery deployment for data centers and utilities is strong, but Tesla cannot make enough batteries, which means it misses a lot of sales.

The panelists agreed that this shortfall is bad for the stock because it means lost sales.

The panel returned to the merger. Musk spoke at length on the SpaceX and Tesla calls, then said he would take no more questions on it. On the SpaceX call he made news, because few knew there were committees examining the potential merger. He said it was a good idea, which raised the question of whether he had dug himself into a hole with shareholders who need someone disinterested.

A Reuters report, as cited by the panel, said insiders involved in the process were unhappy about his vocal comments because they could influence the process and create legal issues. The panel recalled the Solar City precedent: Tesla bought a company started by Musk's cousins, and years of litigation followed, which the panel said eventually worked out for Musk but took years and millions of dollars.

Another side of the issue, which the panel said the New York Times had noted, is that Musk may want the deal done before the 2028 general election. The panel said he currently has a sympathetic White House, with Trump on his side again after the DOGE fallout. A host said Musk wants the transaction done because a Democratic administration would probably scrutinize what would be the biggest corporate merger in history much more carefully, especially since Musk is so unpopular with Democratic leaders.

The panel discussed Musk's posts on Twitter, in which he threatened opponents, saying 'live by the sword, die by the sword' several times in different contexts. The panel said this made explicit what had been implicit, following Trump's own statements that he would go after his enemies. The panelists said a different administration could demand concessions, such as spinning out Starlink for national security reasons, and that Musk would be loath to do that.

The panel noted that corporate America has traditionally avoided taking sides politically and often donated to both parties, partly because of a strong business case for not drawing the attention of the opposing party. A host said this could be a material risk to shareholders. The host added that they would have more reporting later that day, after speaking with Tom Makunda, who teaches leadership at Yale.

Makunda, as the hosts relayed him, said that even leaders with extreme political power in past decades knew when to push and when to dial it back, doing so quietly and keeping everyone happy. The hosts said Musk is doing this very vocally, which they said the country has never really seen. They pointed to John D. Rockefeller before the Standard Oil breakup as having been very careful about whom he aligned with politically and how he pushed it. One host joked that if Rockefeller had Twitter, who knows what would happen.

Senator Ruben Gallego of Arizona invited Musk on X to come talk without waiting for a subpoena. The hosts said that when Congress is taken over, which they said is likely, subpoenas will come. One host said Musk is trying to prevent that, but in doing so is drawing the most attention and raising the likelihood of subpoenas, especially over Starlink and Starshield in Ukraine.

The panel moved to Meta, with a host noting that Tesla is spending less while Meta is spending more. Meta now forecasts capex of $135 to $145 billion for this year, raising the bottom of the range by $5 billion. Meta has spent $86 billion so far, so it must ramp up spending in the fourth quarter.

The company said more than 10 million people had used Muse in its first six weeks. Third-quarter revenue rose 26% and beat forecasts, and fourth-quarter revenue is expected to be above forecast.

Mark Zuckerberg said Muse will probably remain free but will make money through the transactions it facilitates. A host gave the example that Meta takes a cut when Muse books a flight. The hosts said the company has not yet disclosed hard numbers on what that revenue could look like, although there is a sense on the street that Muse is already a big product and will continue to be one. Shares were also down.

The hosts raised open questions: whether people will accept paying a cut to Meta on purchases such as flights, hotels or other online buys, and how this fits with Amazon blocking agents like Muse on its platform. One host said Delta Airlines is reportedly doing the same.

A host flagged a comment from Meta's CFO about Muse. The CFO said Muse's compute intensity means expanding the user base will require significant investment, and that total serving costs are expected to rise. The hosts noted that Meta is not giving granularity on downloads or users, but said the investment appears considerable.

Hosts noted that Reality Labs is still losing money. It could lose as much as $19 billion this year and is still expected to lose $15 billion next year.

Hosts discussed a note from analyst Laura Martin, who has a hold rating on Meta and says it is too early to buy because of the company's heavy AI investment. She also said the stock split implies an equity raise, which is a reason to wait, and that Muse will face competitors.

Martin said investors should add Meta's off-balance-sheet obligations, which she puts at $626 billion as of the end of June, making the investment more expensive when counted.

A host recalled reporting that Anthropic's obligations could be as much as $750 billion over the next seven to ten years, according to the Wall Street Journal. The figure for Meta is $825 billion. The hosts described these as signed future spending commitments that have to happen. One panelist said it was not certain whether all of them would be used, but that the figure came from a filing with the SEC.