Starship Development: The Key to All Long-Term Growth

Bullish investor: Starship is on the cusp of commercial operations, and every major growth initiative depends on it delivering — which it is close to doing.

• SpaceX has now completed 12 Starship flight tests, each progressively advancing the program. The 12th test in May 2026 debuted an entirely new-generation vehicle and booster. SpaceX expects Starship to commence payload delivery in the second half of 2026.
• Crucially, full upper-stage reusability is not required to deploy V3 satellites. SpaceX can begin deploying next-gen V3 Starlink satellites (1 Tbps downlink per satellite, vs. ~50 Gbps for current V2 Mini) and V2 Mobile satellites with even partial Starship capability, providing meaningful near-term Connectivity gains well ahead of full system maturity.
• SpaceX has invested over $15B in Starship. This is not a side bet — the entire capital allocation model is built around it, and management has demonstrated an ability to solve problems that many deemed impossible.

Bearish investor: Starship has a history of delays and failures, and any further setbacks cascade across the entire business.

• Starship R&D expense grew from $1.8B in FY2024 to $3.0B in FY2025, and rose to $930M in Q1 2026 alone. This spending pulled Space Segment Adjusted EBITDA from $1.2B in FY2024 to $653M in FY2025, and into negative territory in Q1 2026 at $(351)M.
• Multiple Starship flight tests — including the 12th — resulted in FAA mishap determinations and return-to-flight conditions. The FAA's current regulations don't even permit return-to-launch-site reentries for Starship, requiring a waiver that is not guaranteed.
• Without Starship achieving full reusability and high cadence, orbital AI compute is economically unviable, V3 satellite deployment stalls, and the central thesis of the entire business falls apart. Starship is also the only vehicle capable of launching V3 satellites — Falcon 9 can't do it.

Starlink ARPU Erosion vs. Subscriber Volume Growth

Bullish investor: The volume math has worked overwhelmingly in Starlink's favor so far, and the model is designed to continue working even as ARPU declines.

• Starlink subscribers grew 99.9% YoY to 8.9M at end-FY2025, and further to 10.3M — a 105% YoY increase — by March 31, 2026. Despite ARPU declining 11.2% in FY2025, Connectivity segment revenue grew 49.8% and Segment Adjusted EBITDA increased 86.2% to $7.2B in FY2025.
• SpaceX is explicit about its strategy: it is prioritizing subscriber growth over ARPU, expecting satellite manufacturing and launch scale economies to offset per-user revenue dilution over time. The V3 satellite (enabled by Starship) will deliver ~20x the downlink capacity per launch compared to Falcon 9, significantly reducing cost-per-bit served.
• Enterprise and government revenue is growing fast alongside consumer. Enterprise and government revenue grew $1.4B in FY2025, led by $1.2B from enterprise mobility (aviation, maritime). No enterprise customer generating over $750K in annual revenue has voluntarily churned since 2023.

Bearish investor: ARPU is declining faster than is sustainable, and the cost offset mechanism (Starship + V3 satellites) hasn't materialized yet.

• Monthly ARPU has fallen from $99 in FY2023 to $91 in FY2024 to $81 in FY2025, and further to $66 in Q1 2026 — a 33% decline in roughly two years, with management explicitly stating it expects ARPU to continue falling.
• As Starlink penetrates its most addressable markets (rural North America, Western Europe, Australia), incremental subscribers increasingly come from lower-income geographies with structurally lower pricing. Eventually, subscriber growth rates will slow as high-ARPU markets saturate, while ARPU keeps falling.
• The V3 cost offset depends entirely on Starship, which is not yet commercially operational. Until it is, every additional subscriber in low-ARPU markets modestly dilutes economics without the benefit of improved unit economics.

Returns on AI Compute Investment vs. Competitors

Bullish investor: SpaceX has unique structural advantages that separate its AI infrastructure from the generic hyperscaler build-out.

• SpaceX claims a dual speed and cost advantage: it brought the first COLOSSUS cluster of ~130MW online in 122 days, and the first COLOSSUS II cluster online in just 91 days — versus an industry benchmark of approximately two years for a 100MW greenfield data center.
• The Anthropic cloud services agreement signed in May 2026 provides $1.25B/month in compute revenue, immediately demonstrating the commercial value of the infrastructure. SpaceX has stated it expects to sign additional third-party compute agreements.
• The long-term thesis — orbital AI compute powered by solar energy, bypassing terrestrial power grid constraints — is a genuinely differentiated strategy that no competitor is positioned to replicate. The orbital AI compute path leverages SpaceX's launch capability as a moat, not just its data center construction skills.

Bearish investor: AI capex is consuming capital at an extraordinary rate, with returns highly uncertain and a long path to profitability.

• AI segment capex was $12.7B in FY2025 and $7.7B in Q1 2026 alone — annualizing above $30B. The AI segment posted an operating loss of $(6.4)B in FY2025, expanding from $(1.6)B in FY2024. Revenue of $3.2B in FY2025 is a fraction of the capital being deployed.
• The Anthropic agreement — SpaceX's main evidence of third-party compute monetization — is terminable by either party on 90 days' notice after an initial three-month period. It is not a committed long-term revenue stream.
• SpaceX acknowledges a "multi-year investment horizon" before AI Segment Adjusted EBITDA turns positive. Meanwhile, Microsoft/Azure, Google Cloud, and AWS are competing with more established enterprise AI customer relationships, larger developer ecosystems, and their own rapidly scaling infrastructure.

Grok and X vs. Entrenched AI and Social Media Competitors

Bullish investor: Grok's X integration is a genuine differentiator, and SpaceX's compute scale gives it a faster model iteration cycle than most competitors.

• Grok's integration with X provides proprietary access to ~350M daily posts — a real-time data stream no competitor can replicate. SpaceX argues this makes Grok uniquely capable for time-sensitive, current-events-heavy use cases.
• Grok has achieved one of the fastest model iteration cycles in the industry, releasing four major model versions in roughly two years and achieving frontier-level performance on the GPQA Diamond scientific reasoning benchmark faster than other leading providers.
• As of March 31, 2026, 117M of 550M MAUs were using Grok's AI features — still early in monetization, with significant room to convert free users to paid subscribers. Grok's paid subscriber base grew from 0.9M SuperGrok/Heavy subscribers at end-2025 to 1.9M by March 31, 2026.

Bearish investor: Grok is competing against deeply entrenched, better-resourced AI providers, and X's advertiser base continues to deteriorate.

• Grok competes directly against OpenAI (backed by Microsoft), Google Gemini, Anthropic (backed by Google and Amazon), and Meta's Llama — all with larger developer ecosystems and more established enterprise relationships. SpaceX is a later entrant in a market where incumbency matters.
• X advertising revenue fell 23% YoY in Q1 2026 to $343M, down from $443M in Q1 2025 — worsening from an already declining trend. X advertising revenue fell 11.5% from FY2023 to FY2024 as advertisers exited following the 2022 acquisition of Twitter, and the slide is accelerating. A declining X weakens the data advantage that supposedly differentiates Grok.
• Only ~6.3M of 550M MAUs were paying subscribers as of March 31, 2026 — a thin monetization base relative to the billions in capital deployed in the AI segment. The commercial AI model market is moving fast, and Grok has yet to establish a durable enterprise customer base.

Debt Burden and Capital Allocation Across Three Capital-Intensive Segments

Bullish investor: The IPO transforms SpaceX's balance sheet, and the Connectivity segment generates enough cash flow to underpin continued investment across all three segments.

• The IPO raised approximately $74.4B in net proceeds, which fundamentally changes the liquidity picture. SpaceX is required to use IPO proceeds to repay the $20B SpaceX Bridge Loan within six months, and will do so with ample capital remaining.
• Connectivity generated $7.2B in Segment Adjusted EBITDA in FY2025 and $2.1B in Q1 2026 — a growing cash engine that funds the other two segments and provides a stable baseline even as AI investment ramps.
• SpaceX has a proven track record of capital allocation: the Space segment turned Adjusted EBITDA positive in 2018 and the Connectivity segment in 2023. Management is executing the same playbook for AI.

Bearish investor: Total debt of $29.1B, combined with capex commitments and three simultaneously money-losing investment programs, creates a fragile financial structure.

• Total debt as of March 31, 2026 was $29.1B, with the $20B SpaceX Bridge Loan maturing on September 2, 2027 (with two three-month extensions). Interest expense was $1.9B in FY2025 and $664M in Q1 2026 — a material ongoing cash drain.
• Total capex was $10.1B in Q1 2026 alone, meaning SpaceX is on pace to spend roughly $40B+ in capex in 2026, well in excess of the $6.8B in operating cash flow generated in FY2025. The company also committed to a $19.6B spectrum acquisition from EchoStar, expected to close in November 2027.
• The company posted a net loss of $(4.9)B in FY2025 and $(4.3)B in Q1 2026, and carries an accumulated deficit of $(41.3)B as of March 31, 2026. If AI returns don't materialize on schedule, or if Connectivity ARPU declines faster than subscriber growth compensates, the cash position could be strained while simultaneously trying to refinance $20B of bridge debt.