In 2025, SpaceX’s Starlink-led connectivity business generated $11.4 billion—apparently the first satellite-broadband operation ever to cross $10 billion in annual revenue—and supplied 61 percent of the company’s total sales. More strikingly, it earned $4.4 billion in operating profit while SpaceX’s rocket division lost $657 million, making internet connectivity the financial engine helping bankroll Starship.

SpaceX’s most profitable disclosed business in 2025 was not carrying customers into orbit. It was selling connectivity through a segment led by Starlink.

The company’s June 2026 prospectus reported $11.387 billion in revenue for Connectivity, alongside $4.423 billion in income from operations. Space, the segment containing Falcon, Dragon and Starship, generated $4.086 billion of revenue and recorded a $657 million operating loss.

Those numbers do not make the rockets incidental. Without Falcon, SpaceX could not have built Starlink at its present scale. They do reveal that the financial relationship has reversed: a launch system originally funded from outside capital and customer missions created a network that can now help pay for the next launch system.

The 61 per cent is straightforward arithmetic

SpaceX’s official prospectus gives 2025 revenue of $18.674 billion for the combined group. Connectivity contributed $11.387 billion, Space $4.086 billion and AI $3.201 billion. Dividing $11.387 billion by $18.674 billion produces 60.98 per cent, which rounds to the 61 per cent in the title.

Connectivity is broader than household Starlink dishes. SpaceX says the segment is primarily driven by Starlink and includes consumer, enterprise, government and mobile connectivity. Revenue grew 49.8 per cent from 2024, when the segment recorded $7.599 billion. It had reported $3.869 billion in 2023.

The progression matters. Starlink-led revenue almost tripled in two years and crossed $10 billion without relying on a single launch or hardware sale. Much of it came from recurring service relationships spread across different types of customer.

Why “apparently the first” needs both words

There is no complete historical register covering every satellite-broadband operation. Some companies were private. Public operators have also changed their segment definitions, bought one another and combined broadband with television distribution, wholesale satellite capacity, defence systems or terrestrial wireless services. A categorical “first ever” would claim more certainty than the records allow.

The disclosed comparisons nevertheless make the threshold look unprecedented. EchoStar, which contains HughesNet, reported $1.456 billion for Broadband and Satellite Services in 2025. Its $15.005 billion group revenue was mostly pay television and wireless, not satellite broadband.

Viasat’s entire company generated about $4.5 billion in its 2025 financial year, including defence technology and other activities. SES reported €3.512 billion of pro forma revenue after treating Intelsat as consolidated for the whole of 2025. Eutelsat’s total for 2024-25 was €1.244 billion, with its OneWeb low-Earth-orbit operation contributing €187 million.

These are not perfectly equivalent businesses or accounting periods. That is the reason for “apparently”. They do show that Starlink-led Connectivity did not edge past a familiar industry record. It reached a scale several times larger than the annual revenue of major incumbent satellite groups.

The $4.4 billion is operating income, not cash flow

Connectivity’s $4.423 billion was income from operations after $6.964 billion of segment costs and expenses. It implies an operating margin of roughly 38.8 per cent. The result more than doubled from $2.006 billion in 2024.

Operating income should not be confused with net income or free cash flow. It excludes interest and tax, and an asset-heavy satellite network still spends large sums on spacecraft, launches, terminals, ground systems and working capital. The prospectus separately reports $7.168 billion of adjusted segment EBITDA, a measure that adds back depreciation, share-based compensation and certain other items.

The same care applies to Space’s $657 million operating loss. That segment reported positive adjusted EBITDA of $653 million. Its operating result included depreciation, share-based compensation and $3.004 billion of research and development. Saying the rocket division “lost $657 million” is accurate only when “operating” remains attached to the number. It is not proof that Falcon launches consumed $657 million in cash.

The expensive item inside Space was Starship

Space revenue increased 7.6 per cent in 2025, while launch-services revenue stayed approximately flat. Cost of revenue fell by $189 million, with greater Falcon reusability reducing depreciation by $240 million. The mature launch system was becoming cheaper to operate on that measure.

Research and development moved the other way. Space R&D rose 63.7 per cent to $3.004 billion. SpaceX attributed the increase primarily to accelerated Starship spending, including production, test launches, engineering and the facilities intended to support future flights.

That makes the segment comparison more informative than a simple story about internet succeeding while rockets fail. Falcon and Dragon generated revenue and deployed Starlink. Starship remained a development programme. The accounts group the mature system and its costly proposed successor together.

Our earlier examination of the filing focused on that segment structure. The sharper point in the newly stated title is the funding loop: Connectivity’s operating income exceeded the Starship-heavy Space segment’s entire research budget by about $1.419 billion.

“Bankrolling” is an inference, but a grounded one

SpaceX does not disclose a bank account into which Starlink subscriptions flow before being assigned to a particular Starship test. The company raises debt and equity, buys equipment, holds cash and allocates capital across a consolidated organisation. Individual dollars cannot be followed from a household terminal to a Raptor engine.

The prospectus does describe the strategy. SpaceX says its Space business first helped fund Connectivity, creating predictable recurring revenue from consumer, enterprise and government customers. It says the company now continues to invest in both segments through the Starship launch platform and expanded Starlink networks. Elsewhere, it says cash from operations enables research and development including Starship and next-generation satellites.

It is therefore reasonable to say Connectivity is helping bankroll Starship, provided that “helping” remains. The $4.423 billion operating profit increased SpaceX’s internal funding capacity while the Space segment absorbed $3.004 billion of research spending. It did not eliminate the company’s need for external capital, nor does it show a one-for-one transfer.

Falcon and Starlink form one industrial system

SpaceX completed 165 Falcon missions in 2025. Only 43 were customer launches; 122 were internal. Many of the internal missions expanded or replenished Starlink. That distinction is easy to miss if launches are counted without asking who bought them.

SpaceX capitalises satellite manufacturing and launch costs inside Connectivity, including an allocation for flight hardware supplied by Space. Those costs then appear through depreciation over the satellites’ useful lives. The two segments are real accounting units, but they are not independent companies trading at arm’s length.

Falcon gives Starlink frequent access to orbit at an internally controlled cost. Starlink gives Falcon a customer large enough to support a cadence no outside launch market has produced. As the constellation has expanded, including the recent crossing of 11,000 satellites in orbit, that vertical integration has become a barrier a broadband competitor cannot solve merely by ordering a few launches.

Recurring service changes the risk profile

A launch provider earns revenue when a mission flies and may wait months or years between a contract award and delivery. Connectivity produces monthly or contracted service revenue after the infrastructure is operating. That revenue can be more predictable, although Starlink still has to replace short-lived satellites, expand capacity and maintain an enormous ground network.

SpaceX said Connectivity subscriber numbers increased 99.9 per cent in 2025 while average revenue per subscriber fell 11.2 per cent, reflecting international expansion and lower-priced plans. Growth came from widening the network rather than simply raising the bill for an established base.

The resulting service revenue gives SpaceX something previous pure launch companies lacked: a large recurring business built on top of the hardware it places in orbit. Starship does not yet provide that kind of dependable commercial flow. It is still being tested, and major requirements such as orbital propellant transfer remain unfinished.

The 61 per cent denominator has an AI complication

SpaceX acquired xAI on 2 February 2026. Because SpaceX, xAI and X were entities under common control, the prospectus retrospectively presents earlier periods as though they had already been combined. The 2025 total therefore includes $3.201 billion of AI revenue and a $6.355 billion AI operating loss even though the legal transaction closed in 2026.

Connectivity’s 61 per cent share belongs to that recast combined group. If only Space and Connectivity are counted, Connectivity supplied about 73.6 per cent of their revenue. Neither calculation is wrong; they answer different questions. The title follows the prospectus’s combined-company presentation.

The AI loss also explains why SpaceX as a whole reported a $2.589 billion operating loss despite Connectivity’s profit. Starlink was the financial engine of the operating businesses, but it was not large enough to offset every cost after the retrospective combination.

Rockets remain the moat; connectivity supplies the engine

SpaceX is still a rocket company in engineering terms. Falcon’s reuse and cadence made Starlink’s scale possible, while Dragon and government launch work remain important businesses. Starship is intended to increase that advantage by carrying more mass more frequently if full reuse can be achieved.

The 2025 accounts change the financial description. Connectivity produced most of the combined group’s revenue, crossed an apparent industry threshold and generated the only positive operating result among SpaceX’s three reportable segments. Space carried the cost of developing Starship and finished with an operating loss.

That is not a story of the internet replacing rockets. It is a story of one SpaceX system financing the next. Falcon built Starlink. Starlink now supplies recurring revenue and operating profit. Those earnings help give SpaceX the time and money to keep working on Starship before Starship can pay its own way.