Space Exploration Technologies Corp. (SpaceX) (SPCX) ☆

Owned GPU data centres (COLOSSUS, COLOSSUS II) used to train Grok, with excess capacity sold as cloud services to third parties.

Neither filing gives a share basis. The company frames itself as an entrant rather than an incumbent, intending "to sell our excess capacity by offering it to a limited number of third parties... potentially positioning us to emerge as a competitor to AI cloud providers such as Coreweave and Nebius as well as hyperscalers". The sourced scale is nameplate compute draw of 1.4 GW at 6/30/26 against 0.4 GW a year earlier, and AI infrastructure revenue up $1,600M in Q2 2026 "as we began to offer cloud services to customers" — a business only months old.

Barriers here are low. The 10-Q states that cloud services agreements, after an initial ramp, "may be terminated by either party upon 90 days notice" and that "a significant portion of our AI infrastructure revenue is concentrated in a small number of customers", leaving near-zero contractual lock-in. Capital intensity is a funding hurdle rather than a customer switching cost: AI absorbed $15,828M of $18,369M segment capex in Q2 2026 and $23,551M of $28,476M in H1, and the filing says it may reallocate this spend away if data centre needs slow.

Reusable capsule flying NASA and private astronauts and cargo to the ISS, launched on Falcon 9.

Leadership here is real but narrower than it looks. The filing states that Falcon 9 paired with Dragon is "the only U.S.-based launch vehicle certified by NASA under the Commercial Crew Program to transport astronauts to and from the International Space Station", and that SpaceX flew "all five U.S. crew and cargo missions to the International Space Station for NASA" in 2025, with 78 crewmembers from 20 countries since 2020 across 19 human spaceflight missions at a 100% success rate. The customer base is essentially one buyer, and Northrop Grumman's Cygnus is named as a cargo competitor.

NASA human-rating certification is a genuine structural barrier and the filing claims exclusivity among U.S. vehicles, with launch barriers of capital, technology, licences and clearances compounding it. The risk factors cut the other way: all U.S. government launch contracts are firm fixed-price with milestone payments, funded by annual Congressional appropriation, and the government can "terminate existing contracts at its convenience and without advance notice". The filing also says SpaceX may prioritise its own payloads over additional U.S. government contracts.

Partially reusable medium and heavy lift orbital launch for commercial, civil, international and U.S. government payloads.

The filing says Falcon 9 "established our position as the leading commercial launch provider". In 2025 Falcon 9 flew 165 launches, "accounting for over half of all global orbital launches in the year while delivering over 80% of mass to orbit", and SpaceX flew 11 of 12 NSSL medium and heavy lift missions. Cadence is easing, however: H1 2026 Falcon launches were 77 versus 81 a year earlier, and the company "expect[s] Falcon 9 launches to decrease over time" as Starship ramps.

The filing lists "substantial capital requirements, advanced technological expertise, regulatory licenses and approvals, and established relationships with government and commercial customers", alongside FAA launch licences. The cost basis is reusability: a first stage has "demonstrated the ability to refly... 34 times" as of 3/31/26, across over 570 booster landings. This is management's own assessment — "we believe that SpaceX holds a meaningful advantage" — not a third-party ranking, and the filing warns "there can be no assurance that we will maintain this position".

Grok model family (consumer subscriptions, API, Business/Enterprise) integrated with the X social platform and its advertising inventory.

No share figure exists in the filing, so this band is a position read rather than a measured share. The filing names OpenAI, Anthropic, Google, Meta, Microsoft, open source providers, Threads, Reddit and TikTok as competitors and concedes that "some of our current and potential competitors, particularly in our AI segment, have greater financial, technical, manufacturing, or other resources than we do". The sourced figures are scale only: roughly 1.3bn supported accounts active over the last twelve months and about 550M MAUs across Grok and X at 3/31/26, of which roughly 117M used Grok's AI features, against about 350M daily posts. That supports a challenger position, not a leading one.

The filing does claim AI entry barriers — "substantial computational and infrastructure requirements, access to large datasets and the ability to attract and retain highly skilled technical talent" — and the X real-time corpus is a genuine proprietary data input. Against that, the AI segment "remains in a relatively early stage of organizational and operational maturity" after the February 2026 xAI acquisition, "increased competition with our AI products could result in price reductions", X advertisers "generally do not have long-term commitments", and advertising revenue fell $59M year over year in Q2 2026 during an ad-platform transition.

Cellular connectivity direct to standard smartphones via dedicated satellite-to-mobile satellites, sold through mobile network operator partners.

Share here is unknown. The filing gives no share, ranking or leadership claim for satellite-to-mobile anywhere; it names AST SpaceMobile, Lynk, Globalstar and Skylo as competitors and says the offering "operates in a highly competitive and evolving market". The only sourced figures are usage rather than share: "approximately 7.4 million monthly unique devices across approximately 30 countries" at 3/31/26, with the service delivered in partnership with mobile network operators.

Connectivity barriers apply — capital, spectrum and orbital resources, licences and customer relationships — and the EchoStar deal adds 65 MHz of U.S. spectrum plus global MSS licences with "very senior ITU priority". But the filing states the offering "may be affected by the pace of technological development, spectrum availability, and the success of our partnerships with mobile carriers", a partner-dependent position, and the next-generation constellation designed to use the acquired spectrum is not yet deployed.

LEO satellite broadband for residential, roam, enterprise and government customers.

Starlink leads satellite broadband, not broadband overall: the filing lists Verizon, Comcast, AT&T, Charter, T-Mobile and others as competitors and concedes service "may be less competitive in certain markets, including dense urban areas". The basis for the band is the filing's statement that "we... deployed the largest LEO constellation in existence", roughly 9,600 Starlink satellites or "approximately 75% of all active maneuverable satellites in orbit" at 3/31/26, and that "Starlink is the sole low-latency network available globally", with low-latency self-defined as under 70 milliseconds. Subscribers were 12.0M versus 6.0M a year earlier per the Q2 10-Q.

The filing cites "significant barriers to entry, including substantial capital requirements, advanced technological capabilities, access to spectrum and orbital resources, regulatory licenses and approvals" plus customer relationships, and SpaceX alone self-launches its constellation while some rival operators are its launch customers. Offsetting that, satellites have three-to-five-year useful lives, "inherently shorter than that of the information technology systems and infrastructure they host", making replacement launches a permanent capital charge.