SPCX

Starship Development

• Completed two successful test flights of Starship V3 in the past 90 days; Flight 13 successfully demonstrated core orbital mission capabilities and heat shield reliability, and the vehicle is being recovered for analysis
• The next flight (Flight 14) will be the first to deploy V3 Starlink satellites to operational orbit, with an attempt to catch both the first and second stages planned as early as the end of this month, targeting full Stage 1 and 2 catch capability by the end of 2026
• Long-term aspiration is to grow annual payload to orbit from ~2,500 tons (current Falcon 9 capacity) to over 1 million tons, and ultimately 10 million tons per year, creating a massive scale advantage over all competitors
• Ongoing investments in production infrastructure, including Raptor engine production, Gigabays construction, and multiple launch pad activation at Starbase and Cape Canaveral to support thousands of annual Starship launches

Starlink Connectivity

• Added a record net 1.7 million consumer subscribers in Q2 2026, up from 1.4 million in Q1, ending the quarter with ~10,200 operational satellites (9,600 broadband satellites) delivering 800 terabits per second of total downlink capacity across 167 global markets
• V3 Starlink satellites are ~10x more capable than V2 satellites, with plans to launch 10x more V3 units, delivering a ~100x total increase in constellation bandwidth; operational deployment of V3 satellites will begin on upcoming Starship flights
• Enterprise and government growth was strong: signed a major agreement with American Airlines, activated service for Southwest, Virgin Atlantic, Iberia and Aer Lingus, and won over $6 billion in new U.S. government contracts in Q2; less than 10% penetrated in the large aviation segment, with massive headroom for growth
• Starlink Mobile: launched new international carrier partnerships with SoftBank, NTT Docomo and Spark New Zealand; gained FCC approval for the 65 megahertz EchoStar spectrum transfer, a foundational competitive advantage; plans to launch next-generation mobile V2 satellites in 2027 and begin commercial service by the end of 2027

AI and Compute Infrastructure

• Reached 1.4 gigawatts of nameplate compute capacity at the end of Q2 2026, up from 1 gigawatt in Q1 and 400 megawatts a year prior
• Released Grok 4.5 in July, which resulted in a 3x increase in token consumption and positive enterprise feedback; Grok 4.6 is planned for release the week after the call, with Grok 4.7 coming in 3-4 weeks, and Grok 5 (which will incorporate all 25 years of SpaceX engineering data to create a leading engineering AI) planned for release by the end of 2026
• Entered into large new cloud compute agreements with third-party customers, including $1.6 billion in incremental revenue in Q2 and an additional $6.7 billion in contracted revenue in the first weeks of Q3; payback on new compute deployments is less than 1 year
• Announced exclusive long-term partnership with NVIDIA for AI hardware, leveraging the Vera Rubin architecture; plans to launch Starmind AI satellites with integrated NVL72 computers starting in 2027, with ground-based deployment of the optimized design planned as well
• The Cursor AI acquisition is nearly complete through all regulatory hurdles, and is expected to close soon

Capital and Balance Sheet

• Completed the IPO in Q2 2026, raising $85.7 billion in net proceeds, followed by a $25 billion investment-grade bond offering; ended the quarter with $100 billion in cash, cash equivalents and marketable securities
• Total Q2 capital expenditures were $18.4 billion, with $15.8 billion allocated to AI compute infrastructure, and the remainder to Starship development, launch infrastructure, and satellite production

• Compute capacity: Targeting over 2 gigawatts of operational compute capacity by the end of 2026; aiming for cumulative projects totaling 20 gigawatts of power and cooling capacity by the end of 2027, with an expected 10-15 gigawatts of operational capacity by the end of 2027
• Revenue: On track to reach $100 billion in annualized revenue run rate (ARR) by the end of 2026, with management noting this target will likely be exceeded; pulled forward the $1 trillion annual revenue target from 2031 to 2030, with a non-zero chance of hitting this target in 2029
• Starship: Flight cadence will increase rapidly, with management expecting at least one flight per day (possibly more) one year from now; the heat shield problem is considered solved, with no remaining technical obstacles to full and rapid reusability; human-rated Starship reliability is expected to be achieved by the end of 2027
• Starlink V3: Will reach the critical mass of ~1,000 V3 satellites needed for widespread service improvements by Q2 2027; management expects Starlink to eventually deliver the majority of global Internet in countries where SpaceX operates within 10 years
• Starlink Mobile: Will begin launching next-generation mobile satellites in 2027, with commercial service starting by the end of 2027
• Starship Human Landing System (HLS): Uncrewed lunar cargo mission will follow the 2027 Artemis III docking mission, with the first crewed lunar landing planned for 2028

1. Space Segment: Revenue grew 29% year-over-year to $962 million, driven by higher volumes of larger customer launches and favorable customer mix. R&D investments in Starship development and launch infrastructure caused an adjusted EBITDA loss of $205 million for the quarter. This segment currently accounts for approximately 12.3% of total company revenue.
2. Connectivity (Starlink) Segment: Revenue reached $4.3 billion, up 66% year-over-year and 32% sequentially, driven by record net subscriber additions and 108% year-over-year growth in enterprise and government revenue. ARPU remained stable at $66 per month. Operating income grew 79% year-over-year to $1.7 billion, with 300 basis points of operating margin expansion. Adjusted EBITDA rose 64% year-over-year to $2.6 billion. This segment accounts for approximately 55.1% of total company revenue.
3. AI Segment: Revenue hit $2.6 billion, up 247% year-over-year and 213% sequentially, driven by new third-party cloud compute services agreements (which contributed $1.6 billion in incremental revenue) and growth in Grok and X subscription revenue. Advertising revenue grew 7% sequentially. The segment narrowed its net operating loss to $1.3 billion and turned adjusted EBITDA positive at $1.1 billion. This segment accounts for approximately 33.3% of total company revenue.

Total company revenue for Q2 2026 was $7.8 billion, up 92% year-over-year, with a net loss of $541 million (an improvement of $467 million year-over-year) and adjusted EBITDA of $3.5 billion, up 191% year-over-year.

• Forward-looking statements are inherently uncertain, and actual results may differ materially due to risks and uncertainties disclosed in SpaceX's SEC filings, including the prospectus and Q2 10-Q
• Starship development and testing carry inherent technical risk; while the heat shield is currently believed to be solved, unexpected issues may still arise during upcoming flight tests and catch attempts
• Large-scale compute expansion depends on GPU supply from NVIDIA, and permitting and infrastructure deployment for large data centers may face delays
• Starlink Mobile's large-scale U.S. market entry depends on successful deployment of a new distributed terrestrial base station model, and may require additional spectrum purchases and capital expenditures beyond current plans
• Global regulatory approval remains a constraint for Starlink expansion in some markets

Q: What is the outlook for enterprise and government Starlink revenue conversion and growth over the next few years? / A: Management is highly bullish on this segment, noting over $6 billion in new government contracts were secured in Q2 alone, and enterprise customers are extremely sticky with zero customer churn to date. Starlink currently has less than 10% penetration in the large aviation market, with massive untapped opportunity in maritime as well. Most enterprise customers still hold outdated perceptions of older, less reliable Starlink service, so SpaceX is expanding its enterprise sales team to demonstrate current high reliability and uptime, which will enable conversion to primary provider status. Management expects enterprise and government revenue will ultimately exceed consumer Starlink revenue.

Q: What is your confidence in reaching 10+ gigawatts of compute by end of 2027, and how does SpaceX's rocket/satellite expertise help accelerate this buildout? / A: SpaceX is targeting cumulative projects totaling 20 gigawatts of power/cooling by end of 2027, and expects ~10-15 gigawatts of operational capacity even if some projects are delayed. SpaceX has secured agreements for a very significant share of NVIDIA's GPU production for 2027. SpaceX's core expertise building reliably reusable rockets (a far harder engineering challenge than data centers) yields massive efficiency advantages when applied to data center construction, and the company already has excess power and cooling infrastructure built relative to GPU availability. Management expects the compute supply-demand imbalance will persist, as demand grows ~200% annually while memory supply only grows ~20% annually, supporting sustained strong pricing.

Q: What is the CapEx outlook going forward, and how does the economics of current AI investments compare to historical SpaceX investments? / A: Management expects the next two quarters of 2026 will have similar CapEx levels to Q2 2026's $18.4 billion. Unlike historical long-lived SpaceX assets like launch sites that generate returns over decades, current AI compute deployments have a payback period of less than 1 year, meaning capital is recycled very quickly into new capacity. SpaceX remains focused on maintaining capital efficiency and strong returns on investment across all segments, while preserving its strong investment-grade balance sheet.

Q: How large is the Starlink Mobile opportunity, and what is the base case for its U.S. market entry and CapEx requirements? / A: The EchoStar spectrum purchase includes terrestrial rights, so SpaceX intends to build out a full terrestrial mobile network, not just operate as an MVNO. Instead of requiring massive upfront capital expenditure for large centralized base stations and expensive low-band spectrum auctions, SpaceX plans to deploy small distributed femtocells integrated with existing Starlink consumer dishes on homes and businesses across the U.S., which management expects will be far more capital efficient. This approach will deliver better coverage with no dead zones, positioning Starlink Mobile to capture significant share from existing U.S. carriers that generate ~$600 billion in annual domestic revenue.