SpaceX Outgrows Its Earnings

If You Only Read One Thing

SpaceX’s fastest-growing business still contributes surprisingly little operating profit. Its earnings release separates the AI revenue surge from the satellite business generating most earnings. Netflix faces a different conversion problem: a huge audience has not delivered its expected advertising sales. Together, the results locate the work between owning a valuable asset and collecting a durable return from it.

SpaceX’s Revenue Engine Is Not Its Profit Engine

SpaceX now has two different centers of gravity: AI drives sales, while connectivity supplies most operating profit. That distinction matters more than the headline earnings beat because shareholders own the capital demands of both businesses.

Tuesday’s earnings release put quarterly revenue at $14.575 billion, up 161%. The AI business, now reported as Super Intelligence, contributed $7.560 billion of sales and $431 million of operating income. Connectivity produced $5.248 billion of revenue and $2.367 billion of operating income. Against consolidated operating income of $2.791 billion, connectivity supplied about 85%. These are operating results, before financing and taxes, rather than cash-flow measures.

The company’s largest investment program sits elsewhere. AI accounted for $16.444 billion of the quarter’s $19.593 billion in capital expenditure. Equipment spending can support several years of revenue, so comparing one quarter’s spending with one quarter’s profit cannot establish project returns. It does establish where the next payback test lies.

This advances our October 9 discussion of Starlink’s prospective mobile expansion. The new information is actual segment economics: an established distribution business sits beside a much more capital-hungry growth engine. The standing capacity thesis still applies, but growing demand does not tell us whether the capacity owner or its equipment suppliers retain the better return.

The strongest bullish argument is that infrastructure costs precede utilization. Once customers fill installed machines, incremental sales can become more profitable. The quarter’s positive AI operating income supports that possibility. The counterweight is continuing expansion: a profitable cohort of machines can coexist with a company that must keep raising money for the next cohort.

Merger exploration adds another allocation question. Musk said SpaceX and Tesla have established independent-director committees, according to Axios’s report of the earnings call. There is no agreed transaction. Combining businesses could coordinate investment; the exchange terms would determine which shareholders finance it and receive the gains.

My medium-confidence investment read over the next 12–24 months favors suppliers paid as capacity is installed over assuming the entire AI revenue increase becomes shareholder cash. Within SpaceX, connectivity provides the stronger demonstrated operating economics. This read weakens if AI operating margins expand while capital intensity falls. The next quarterly segment table provides a concrete test: AI operating income as a share of AI revenue, currently about 5.7%.

Netflix Has to Earn the Advertising Budget

Netflix’s advertising disappointment exposes a second customer relationship it still has to build. Winning the viewer gives Netflix inventory to sell. Winning the advertiser requires proof that buying that inventory produces results.

The October 20 shareholder letter lowered expected annual advertising revenue from roughly $3 billion to $2.7 billion, citing slower delivery of advertising capabilities. That is a 10% reduction. Yet subscription revenue growth exceeded management’s expectations, and the updated $51.1 billion annual revenue forecast sits only 0.2% below the previous midpoint. The miss is concentrated in the new monetization engine.

The same day’s product announcement helps explain the constraint. Netflix introduced a conversion-tracking pixel and enhanced its conversion interface, allowing advertisers to connect customer actions with campaigns. It also said The Trade Desk audience targeting would arrive next month. The Trade Desk separately confirmed that advertisers will be able to bring their own audiences and third-party data into Netflix purchases through its platform.

That integration changes how Netflix competes for money. An advertiser already buying across several services can carry an existing customer segment into Netflix rather than build a separate campaign process. Measurement then helps determine whether another dollar belongs there. These are prerequisites for repeat budget allocation, not merely conveniences around a large audience.

The important long-term pattern is that distribution ownership has limits. Netflix controls the viewing relationship, but advertisers bring their own data, purchasing systems and standards of proof. Some value therefore remains with intermediaries that make spending comparable across publishers. This is the distribution argument familiar from AI, applied to a separate market; it does not establish a new AI-thesis data point.

The counterargument is straightforward: an engineering delay can be temporary. A smaller advertising forecast does not prove viewers reject the ad tier or that advertisers reject Netflix. Conversely, shipping the promised tools will not by itself prove additional sales. Buyers must use them, see acceptable results and return.

With medium confidence, Netflix’s existing subscription franchise is better supported over the next year than an advertising-led acceleration. The Trade Desk is positioned to benefit from easier Netflix buying, although incremental revenue and retained fees remain undisclosed. Strong ad growth after the integrations would weaken that cautious read. The clearest financial test is whether Netflix’s next annual outlook shows advertising growth accelerating after the current capability delays.

The Contrarian Take

Everyone says: A platform that owns both the infrastructure and the customer should keep more of the economics.

Here’s why that’s incomplete: Ownership also combines different investment clocks inside one shareholder claim. SpaceX can own a profitable satellite relationship and still face a much larger AI construction bill; integration does not make those spending schedules match. Netflix’s partner announcement shows why reaching outside the company can help: advertisers can use data and buying tools they already possess. The relevant test is whether ownership improves the return on the next investment enough to justify its additional cost. Control has economic value, but there is no reason to assume every adjacent activity belongs on the same balance sheet.

Under the Radar

SpaceX’s adjusted profit changes the apparent leader. The filing’s AI segment shows $4.953 billion of adjusted EBITDA versus $3.499 billion for connectivity, reversing the operating-income ranking. EBITDA means earnings before interest, taxes, depreciation and amortization; an adjusted version excludes additional specified expenses. Both measures can inform analysis, but using the adjusted ranking alone gives less weight to asset consumption precisely where investment is largest. The ranking depends on which costs remain in the comparison.
Netflix is making the advertiser’s data portable. The Trade Desk says its November integration will support both advertisers’ own audiences and third-party segments. That lets a buyer carry a campaign definition across publishers. My inference is that easier entry into Netflix could also make Netflix easier to compare with rivals: portability reduces buying friction while strengthening the buyer’s ability to redirect spending. The announced integration establishes neither exclusive access nor guaranteed budget growth.

Quick Takes

PHP’s emergency release creates a coordination deadline. The PHP team scheduled fixes for two critical vulnerabilities, both rated 9.2, for October 21 at 14:00 UTC. Its notice reserves affected-version details for release; broad exploit claims should not substitute for that scope. The business exposure is the interval between a shared upstream fix and deployment across individual sites, making maintenance ownership consequential. (Source)

Browser protection can arrive before the rest of the software estate. Mozilla says it blocked eight compromised eMudhra certificates for standard Firefox configurations within three hours of receiving the October 17 report. Its October 20 library update extends protection to applications using its trust store without the browser’s update channel. A common trust foundation can therefore produce different recovery times across products. (Source)

Grid resilience funding is a procurement opportunity before a price result. The Energy Department announced approximately $967 million in additional state and tribal formula-grant funding on October 20. Local projects can create demand for grid upgrades and installation work. Over the next year, awards and procurement matter more for suppliers than the department’s promised lower bills; allocations alone establish neither completed infrastructure nor measurable household savings. (Source)

The Thread

The underappreciated strategic choice is which activities need common ownership and which need a working connection. A SpaceX–Tesla merger would put more investment decisions inside one corporate boundary. Netflix’s advertising expansion relies on connections to systems controlled by outsiders. My inference is that the right boundary depends on where coordination failures are most expensive, rather than on a general preference for building everything internally.

This creates a specific valuation problem. Integration deserves a premium only when joint decisions improve results enough to offset the capital and organizational obligations absorbed. A partnership deserves credit when it supplies the same coordination more cheaply. Netflix’s buyer integrations offer a practical comparison: the service can become easier to monetize without Netflix owning the advertiser’s entire purchasing system.

The ownership case would strengthen if SpaceX eventually documented savings or revenue gains that arm’s-length agreements could not provide. The partnership case would weaken if Netflix’s external integrations repeatedly delayed monetization despite ready customer demand. Those are tests of organizational design, beyond whether either company reports another quarter of growth.

Predictions

I predict: The Trade Desk will make the announced first- and third-party audience targeting for Netflix available through Kokai by November 30. This tests partner delivery, separately from whether Netflix earns more advertising revenue. (Confidence: medium; Check by: 2026-11-30)

2026-10-21 · 03:15 ET