Cursor data and Grok's cost-performance strengthen the enterprise AI pathway; JPMorgan maintains its Overweight rating on SpaceX

Overview

The report focuses on SpaceX's latest AI developments and views the Cursor acquisition, Grok model iterations, and Grok Bot as key steps in the company's expansion into the enterprise AI market. JPMorgan believes AI will gradually shift from the current emphasis on high-end compute transactions toward Grok's enterprise commercialization, while reiterating that launch, Starlink connectivity, and compute infrastructure collectively form the foundation for long-term growth.

Core views

SpaceX completed its acquisition of Cursor on August 14. Cursor had approximately $4 billion in ARR as of June 2026, with about 75% coming from enterprise customers; the report believes this customer base could simplify SpaceX's sales pathway into the enterprise AI market, while Cursor's accumulated data could be used for supplemental model training. The report states that Cursor data has already been used for supplemental Grok training and has produced visible improvements in recent model performance. Combined with approximately 25 years of SpaceX's proprietary internal company data, future models' coding and real-world engineering capabilities are expected to strengthen further. Grok 4.6 was announced on August 12, approximately one month after Grok 4.5 was released on July 16. It achieved a score of 61 on the Artificial Analysis Intelligence Index, up 5 points from the prior version, tying GPT 5.6-Sol and trailing only Anthropic's Claude Fable 5 and Claude Opus 5. The model focuses on long-running agents and was trained using high-quality data covering advanced technical and engineering content, as well as multiple categories of agent reinforcement-learning tasks. Pricing starts at $2 and $6 per million input and output tokens, respectively, which the report believes places it on the Pareto frontier between intelligence and per-task cost. Although Grok's current adoption still trails peers, frontier-level performance and materially lower costs could help it expand penetration among developers and enterprise customers. SpaceX launched an early test version of Grok Bot on August 11. This is a group of continuously online agents capable of performing tasks within applications and websites already used by employees; users can create persistent Bots, assign roles, and delegate work to them as they would to colleagues. These Bots can directly operate application interfaces, retain memory across tasks, and collaborate with one another. SpaceX is already using the product internally across engineering, growth, and marketing departments, while desktop and iOS testing is currently open to premium subscribers of SuperGrok Heavy, Cursor Ultra, and Cursor Teams. The report says early user feedback has been positive and views the product as an important vehicle for SpaceX to participate in the rapidly growing enterprise AI agent market. The model release cadence remains another central theme. The report expects SpaceX to maintain a monthly cadence of new model launches throughout the second half of 2026 and through year-end. Initial training of Grok 4.7 has been completed, with release expected in early September; the company describes it as superior to 4.6 in every respect except slightly slower serving speed, while also offering greater token efficiency. SpaceX is using its unique corpus of company data for supplemental training to strengthen engineering capabilities. Grok 5 is expected to launch before year-end, and the report believes its intelligence and overall capabilities will improve more significantly. The model is believed to have been trained on the Colossus 2 Blackwell supercluster, is expected to have more than 6 trillion parameters, and will focus on improving coding and engineering, agent behavior, multi-step reasoning, and video understanding. Regarding commercialization, recent AI revenue primarily comes from high-end compute transactions priced at $30 to $50 per watt; as model capabilities and products mature, the report expects Grok commercialization, particularly contributions from enterprise customers, to become an increasingly important growth driver. This view is also incorporated into SpaceX's broader compute roadmap: terrestrial compute capacity is expected to increase from 2.2GW in 2026 to 8.6GW by the end of 2028; beginning in 2029, the company is expected to develop orbital compute capacity, reaching approximately 75GW by the end of 2031, which the report believes could cost significantly less than terrestrial construction. The report's long-term investment thesis is not limited to AI. SpaceX has completed approximately 670 orbital launches, with a success rate exceeding 99%, and has carried more than 80% of global mass to orbit since 2023. Starship is viewed as a critical enabling platform, expected to deliver a 10-fold cost improvement and approximately 4 times greater payload capacity relative to Falcon 9, while structurally reducing launch costs by approximately 80% or more compared with traditional approaches. The report expects launches to increase from a small number in 2026 to approximately 5,000 in 2031. In connectivity, Starlink is already the largest low-Earth-orbit constellation, with broadband users expected to increase from 9 million in 2025 to more than 95 million in 2030; mobile services also offer long-term optionality, with international markets potentially first converting existing revenue-sharing agreements through an MVNO model. Based on this, JPMorgan believes the company's combined addressable market across launch, connectivity, and AI exceeds $28 trillion. Regarding valuation, the December 2027 price target of $240 corresponds to approximately 34 times estimated 2028 GAAP EPS of $7.02, above the approximately 20 times median for the large-cap technology peer group of “Mag7+PLTR.” The report attributes this premium to SpaceX's market leadership and differentiated growth and margins across launch, connectivity, and AI: total revenue is expected to increase from $19 billion in 2025 to $196 billion in 2028, representing a 119% compound annual growth rate versus a 26% peer average; GAAP operating margin is expected to rise from -14% in 2025 to approximately 44% in 2028, above the peer average of approximately 38% in 2028. This growth also depends on highly intensive execution and capital investment. The report expects upfront investment in Starship, Starlink, and compute to generate more than $300 billion in negative free cash flow and debt financing needs from 2026 to 2030, with free cash flow turning positive only in 2031; recent transactions have raised more than $85 billion in IPO proceeds and $25 billion in debt. Starship delays, AI power and chip constraints, competition, regulation, and key-person risk could all disrupt the expected trajectory. In addition, Elon Musk holds approximately 82% of the voting power, and concentrated control creates governance and leadership succession risks. In the near term, investors should also monitor the lock-up expiration of nearly 370 million shares on September 9 and 10, which could increase the public float by 20%.

Analysis framework

The report first analyzes how the Cursor acquisition supplements enterprise customer channels and training data, then compares Grok 4.6's performance and cost using the Artificial Analysis Intelligence Index, peer rankings, and token pricing. It subsequently examines Grok Bot and the product roadmaps for Grok 4.7 and Grok 5, connecting these developments to enterprise AI commercialization. Finally, the report incorporates AI into the long-term operating model for launch, Starlink, and compute infrastructure, determines the price target through revenue, margins, EPS, and peer P/E multiples, and assesses capital consumption, execution, supply, regulatory, and governance risks.

Methodology notes

• Valuation methodologyPE/PEG valuation

  Relative P/E valuation based on forward GAAP EPS

  The report multiplies estimated 2028 GAAP EPS of $7.02 by a P/E multiple of approximately 34 times to derive a December 2027 price target of $240 and compares it with the approximately 20 times median P/E multiple of large-cap technology peers, attributing the valuation premium to growth, margins, and market leadership.

• Company fundamentals and financial frameworkFree cash flow analysis

  Upfront capital investment and the pathway to positive free cash flow

  The report estimates that construction of Starship, Starlink, and compute infrastructure will generate more than $300 billion in negative free cash flow and debt financing needs from 2026 to 2030, and evaluates capital intensity and financing risk based on the expectation that free cash flow will turn positive in 2031.

• (Method outside the vocabulary)

  Pareto frontier of intelligence and per-task cost

  The report compares both model intelligence scores and usage costs; being on the Pareto frontier means it is difficult to find a cheaper model at a given intelligence level or a more capable model at the same cost, illustrating Grok 4.6's cost-performance and potential adoption advantages.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• SpaceX (SPCX.US)

  The report believes the company can integrate its launch capabilities, Starlink connectivity network, Cursor enterprise channel, and Grok models to jointly address a market exceeding $28 trillion.

  Strengths

  Approximately 670 orbital launches, a success rate above 99%, a leading share of mass delivered to orbit, Starlink's scale, vertical integration, engineering culture, and Grok 4.6's performance-cost advantage.

  Weaknesses

  Grok adoption still trails peers, while long-term plans depend heavily on Starship, advanced chips, power supply, and sustained large-scale capital investment.

  Comparison

  The price target applies approximately 34 times estimated 2028 GAAP EPS, above the approximately 20 times median P/E multiple for Mag7 plus PLTR; the report attributes the premium to a 119% revenue CAGR and an approximately 44% operating margin in 2028.

  Risks

  Key risks include Starship delays, AI supply bottlenecks, intensifying competition, prolonged free cash flow consumption, regulatory restrictions, and Elon Musk's concentrated control.

Key data

• Cursor acquisition completion dateAugust 14, 2026SpaceX formally completed the acquisition
• Cursor ARRApproximately $4BAs of June 2026, approximately 75% came from enterprise customers
• Grok 4.6 Intelligence Index61Up 5 points from Grok 4.5, tying GPT 5.6-Sol
• Grok 4.6 pricing$2/$6 per million input/output tokensThe report believes it is on the Pareto frontier of intelligence and per-task cost
• Grok 5 scale6T+ parametersExpected to launch before year-end 2026
• Recent high-end compute transaction pricing$30-$50/wattThe primary source of current AI revenue
• SpaceX internal data accumulationApproximately 25 yearsExpected to enhance model coding and real-world engineering capabilities
• Cumulative orbital launchesApproximately 670Success rate exceeds 99%
• Share of mass delivered to orbit80%+Since 2023
• Starship improvement relative to Falcon 910-fold cost improvement and approximately 4 times greater payloadA core assumption in the report's long-term launch thesis
• Projected launchesFrom a small number in 2026 to approximately 5,000 in 2031Dependent on Starship becoming rapidly and fully reusable
• Starlink broadband usersFrom 9M in 2025 to 95M+ in 2030Connectivity business growth forecast
• Terrestrial compute capacityFrom 2.2GW in 2026E to 8.6GW by year-end 2028AI infrastructure expansion forecast
• Orbital compute capacityApproximately 75GW by year-end 2031Expected to advance beginning in 2029
• Combined addressable market$28T+Covering launch, connectivity, and AI opportunities
• Price target valuationApproximately 34 times 2028E GAAP EPS of $7.02Corresponding to a December 2027 price target of $240
• Revenue forecastFrom $19B in 2025 to $196B in 2028E119% compound annual growth rate versus a 26% peer average
• GAAP operating marginFrom -14% in 2025 to approximately 44% in 2028EPeer average of approximately 38% in 2028E
• Projected negative free cash flow and debt fundingMore than $300B from 2026-2030Positive free cash flow is not expected until 2031
• Recent financing$85B+ IPO and $25B in debtTo support capital-intensive expansion
• Lock-up expirationNearly 370M sharesUnlocking on September 9 and 10, potentially increasing the public float by 20%
• Elon Musk voting powerApproximately 82%Creates concentrated control and key-person risk

Impact & implications

The report believes Cursor brings both an enterprise customer channel and training data to SpaceX, while Grok 4.6 demonstrates the company's ability to improve model performance rapidly while maintaining a clear cost advantage. If subsequent models, agent products, and enterprise sales progress as planned, the AI revenue mix could gradually expand from high-end compute transactions to Grok enterprise commercialization. Longer-term valuation still depends on Starship reducing launch costs, Starlink expanding its user base, and terrestrial and orbital compute infrastructure being built on schedule, while substantial capital requirements and the September lock-up expiration represent near-term constraints.

Risks

• Rapid and full reusability of Starship is a core prerequisite for long-term growth; technical delays, setbacks, or regulatory obstacles could simultaneously weigh on multiple businesses.
• Power and advanced chip supply bottlenecks could delay constellation and compute infrastructure construction, slow the launch cadence, and depress gross margins, while also introducing geopolitical and trade risks.
• The company must sustain highly intensive execution across multiple large-scale, technologically advanced projects; competition from Blue Origin, AMZN, and international participants could affect pricing, win rates, and R&D spending.
• Upfront investment in Starship, Starlink, and compute is expected to generate more than $300 billion in negative free cash flow and debt funding needs from 2026 to 2030; cost overruns or expansion delays could prolong cash burn and increase financing requirements.
• Export controls, ITAR, spectrum allocation, and environmental reviews could slow launch activity, restrict market access, and increase uncertainty around project timelines.
• Elon Musk's approximately 82% voting power creates governance, key-person, and leadership succession risks, while controlled-company status means board independence is relatively limited.

What to watch

• Monitor whether Grok 4.7, expected to launch in early September, can deliver broad performance improvements while increasing token efficiency.
• Monitor whether Grok 5 can launch before year-end 2026 and the extent of its improvements in coding, engineering, agents, multi-step reasoning, and video understanding.
• Monitor feedback from Grok Bot testing and its progress in expanding from premium subscribers into broader enterprise workflows.
• Monitor whether Cursor's enterprise customer channel and training data can translate into improved Grok adoption and enterprise commercialization.
• Monitor the lock-up expiration of nearly 370 million shares on September 9 and 10 and the impact of a potential 20% increase in the public float.
• Monitor whether SpaceX can maintain its plan to release new models every month in the second half of 2026.