Why Anthropic's $120.5 Billion SpaceX Compute Deal Puts Tesla's AI Infrastructure in Focus

Anthropic's disclosed AI compute agreements with SpaceX of up to $120.5 billion frame Tesla's energy storage and Terafab chip manufacturing projects as critical supply-chain nodes in the AI buildout, even as the company's FY2025 capital expenditure of $8.53 billion ranks last among Slicast's hyperscaler peer group by intensity.

Anthropic's disclosure that its AI compute agreements with SpaceX could reach up to $120.5 billion places renewed attention on the supply-chain web that SpaceX anchors — one in which Tesla occupies several distinct roles: energy storage vendor, aspiring chip manufacturer, and, according to its latest regulatory filings, a company now describing its core mission as bringing "SI" rather than simply artificial intelligence into the real world.

Tesla shares closed at $391.62 on October 26, a gain of 5.2% on the session. No published report reviewed by Slicast identifies the specific driver of that move. The session follows a busy earnings period: Tesla filed an 8-K on October 21, 2026, disclosing Q3 2026 financial results, and lodged the accompanying 10-Q on October 22, 2026, which reaffirmed focus on FSD (Supervised), Robotaxi, and Optimus under the company's "SI" framing. An earlier October 2, 2026, filing reported that Tesla produced over 464,000 vehicles, delivered over 486,000, and deployed 13.7 GWh of energy storage products in the third quarter — a deployment figure that increasingly defines Tesla's relevance to the AI infrastructure market.

The energy connection is concrete and growing. SpaceX reportedly spent $329 million on Tesla Megapack systems in 2026 to power xAI's data centers, and Tesla completed what reports describe as the largest grid-scale battery facility in the United States at xAI's Colossus site in Memphis. In June 2026, Tesla, Sunrun, and Renew Home announced a 16-gigawatt virtual power plant targeting data center demand. The more ambitious bet is Terafab: in August 2026, SpaceX and Tesla announced plans for an AI chip manufacturing complex in Texas carrying a reported initial investment of $16.8 billion and a footprint of up to 100 million square feet — which Musk described as 50 times the size of the Pentagon. Musk has also stated that ASML and other semiconductor equipment suppliers are accelerating deliveries to the site, and has confirmed that Tesla and SpaceX, not TSMC, will build and operate the facility. Nvidia's September 2026 inclusion of Tesla among the first partners in its DSX Ready AI factory certification program added external validation to that positioning.

Tesla's balance sheet picture is more mixed. Slicast's own XBRL compilation shows FY2025 capital expenditure at $8.53 billion, a 24.8% decline from FY2024's $11.34 billion, representing 9% of FY2025 revenue of $94.83 billion. That intensity places Tesla seventh out of seven companies in Slicast's hyperscaler peer group and sixth out of six by absolute capex — a profile consistent with a company managing near-term cash discipline while constructing a longer-dated infrastructure ambition. Reports from October 2026 indicate Tesla has also trimmed its 2026 capital expenditure forecast, though specific detail and independent confirmation remain limited. To fund what its filings describe as an expanding infrastructure program, Tesla entered a $20.0 billion senior unsecured delayed-draw term loan facility on September 29, 2026, with Citibank as lead.

The Terafab thesis carries execution risk at nearly every layer. A $16.5 billion deal tied to Samsung Foundry has reportedly been complicated by yield problems at the manufacturer's advanced nodes, according to an August 2026 Slicast report. Tesla's AI5 chip — which taped out at Samsung on a 2nm-class node following an earlier TSMC tape-out — leaves the company operating a dual-foundry strategy whose coordination and cost exposure are difficult to assess externally. A related talent signal: DensityAI, a chip startup founded by former Tesla Dojo engineers, was reported at a $10 billion valuation in September 2026, having raised hundreds of millions of dollars to develop an accelerator with 3D DRAM stacking — suggesting that part of the expertise underpinning Tesla's chip ambitions has been recapitalized independently. Self-operating a facility described as 50 times the size of the Pentagon, while simultaneously running an automotive and energy business at scale, represents an organizational challenge without clear precedent in the semiconductor industry.

Three signals are worth tracking closely in coming quarters. First, the pace of ASML equipment deliveries and construction milestones at Terafab will indicate whether the project is advancing on a competitive schedule; SpaceX has stated it can build a small AI data center in approximately five months, though the Terafab scale is orders of magnitude larger. Second, any material draw on Tesla's $20 billion credit facility and a disclosed use of proceeds would clarify how Terafab financing is being handled relative to core automotive and energy capital requirements. Third, resolution of the Samsung yield situation will determine whether the AI5 chip reaches volume production on the current dual-foundry path or requires a more substantial manufacturing reset. Anthropic's commitment to SpaceX establishes AI compute demand at a scale that is difficult to dismiss; the open question for Tesla is whether its supply-side infrastructure — ambitious in design, but trailing every peer in current capital intensity — can be assembled on a timeline and at a cost its balance sheet can sustain.