Elon Musk argued that NASA should repurpose its Space Launch System workforce to other parts of the economy. The comment follows renewed scrutiny over the multibillion-dollar cost of the agency's Moon rocket program.

Cost Scrutiny and Legacy Mandates

Musk posted on X on Tuesday, replying to criticism that the SLS program has been sustained primarily to preserve legacy aerospace jobs. He stated it would be better to move these workers elsewhere.

The Government Accountability Office noted in 2012 that the 2010 NASA Authorization Act directed the agency to leverage Space Shuttle and Constellation investments for SLS. The NASA Office of Inspector General later said this mandate led to the retention of Shuttle-era systems, infrastructure, and contracts.

Cost estimates highlight the scale of the expenditure. The NASA OIG estimated in 2021 that flying one SLS/Orion system through Artemis IV would average about $4.1 billion per launch. This figure covers the rocket, Orion spacecraft, European Service Module, and ground operations. Production of the SLS alone was estimated at about $2.2 billion per mission.

The NASA OIG stated that the agency underestimated the complexity of adapting heritage engines and boosters. The GAO identified Boeing Co.’s core-stage contract as a cost-plus-award-fee arrangement.

Commercial Shift and Payload Comparison

Policy is shifting toward commercial systems. The Trump administration’s fiscal 2026 NASA budget proposal calls for an "orderly phase out" of SLS after Artemis III. The plan includes using commercial transportation services for later Artemis missions.

Discussions often contrast SLS with Space Exploration Technologies Corp.’s reusable Falcon Heavy. SpaceX states Falcon Heavy can carry 63.8 metric tons to low Earth orbit. NASA lists SLS Block 1 at 95 metric tons to low Earth orbit and more than 27 metric tons toward the Moon. The vehicles are not directly equivalent due to different mission profiles.

What the Numbers Show

The cost structure reveals a significant dependency on non-rocket elements. With total system costs estimated at $4.1 billion and SLS production at $2.2 billion, ground operations, the Orion spacecraft, and the European Service Module account for the remaining $1.9 billion per launch. This indicates that reducing SLS production costs alone would not eliminate the high per-mission expenditure.

Mission Timeline

NASA continues to press ahead with current plans. Artemis II carried astronauts around the Moon in April. The agency is assembling the Artemis III vehicle for a 2027 crewed demonstration mission in low Earth orbit. A planned Artemis IV lunar landing is scheduled for 2028.

Space Exploration Technologies Corp. (NASDAQ: SPCX) has begun monetizing its AI infrastructure by leasing computing capacity to external customers, securing a deal with Anthropic worth approximately $1.25 billion a month.

The agreement grants the Claude maker access to the full capacity of SpaceX’s Colossus data center. This move signals a strategic pivot from treating GPUs and data centers solely as internal assets to operating them as a standalone business line.

SpaceX’s Compute Business

SpaceX is effectively entering competition with cloud providers by renting out infrastructure rather than just building models. The company has signed high-profile compute agreements that leverage its existing hardware for revenue generation.

This strategy reflects a broader shift in the AI economy where infrastructure ownership becomes a primary value driver. SpaceX is positioning itself not only as an AI developer but also as a provider of the underlying power and compute resources.

Meta’s Next AI Opportunity

Meta Platforms, Inc. (NASDAQ: META) appears to be testing a similar path. In July, the social media giant was in early talks to lease up to $10 billion worth of computing capacity to Anthropic over two years.

While discussions remain preliminary and may not result in an agreement, such a deal would mark a significant step toward commercializing Meta’s AI infrastructure. Meta currently plans to deploy 14 gigawatts of computing capacity next year alongside its in-house AI chips.

What the Numbers Show

The scale of these potential contracts highlights a divergence in AI capital allocation. While traditional cloud providers compete on software services, companies like SpaceX and Meta are leveraging balance sheet strength to build physical infrastructure first. The reported $10 billion potential deal for Meta represents a substantial monetization avenue if executed, suggesting that excess capacity management is becoming a critical component of AI strategy.

Industry Context

Broadcom Inc. (NASDAQ: AVGO) CEO Hock Tan recently acknowledged that the pace at which chips are deployed into powered data centers remains a key consideration for forecasts. This suggests the industry constraint is shifting from demand for AI chips to the availability of powered infrastructure.

Companies that own large-scale data centers, GPUs, and power capacity may occupy one of the most profitable positions in the AI ecosystem. The trend indicates that deep-pocketed firms are building massive infrastructure with the intent to monetize excess capacity later.