Meta Cuts Claude, SpaceX Rebrands AI Unit: Oct. 5

Two headline stories dominate today’s AI investment tape: Meta and Microsoft are pulling back from Anthropic’s Claude platform even after heavy spending, and SpaceX is riding a sharp AI-driven rally while rebranding its artificial intelligence unit.

Key points

• Meta reportedly halved employee engagement with Anthropic’s Claude Code after spending $105M on the platform, signaling a pivot toward in-house AI development.
• Microsoft has also scaled back its use of Claude, despite being among Anthropic’s largest enterprise customers.
• SpaceX shares have surged roughly 60% from their early August low, returning Elon Musk to trillionaire status.
• SpaceX plans to rename its AI unit from SpaceXAI to SpaceXSI, adopting a “super intelligence” branding consistent with terminology favored by the Trump administration.
• AM Intelligence, backed by Greenko promoters, ordered 20,000 additional Nvidia Rubin GPUs, scaling its fleet from 9,000 to 29,000 GPUs as demand for AI infrastructure in India accelerates.

What does Meta pulling back from Claude actually signal?

The numbers here are worth sitting with. Meta spent $105 million on Anthropic’s Claude Code, then watched employee usage drop by half. That is a significant efficiency question for any enterprise evaluating third-party AI platforms: does the spend translate to durable adoption?

Both Meta and Microsoft are now reducing their Claude reliance in favor of building or deploying internal AI capabilities. For Meta, the strategic logic is straightforward. The company has invested heavily in its own Llama model family and has the engineering scale to run proprietary coding tools. Paying a competitor $105M for a product that employees then use less and less is hard to justify internally, regardless of Claude’s technical quality.

For Anthropic, losing or seeing reduced spend from two of the largest technology companies in the world is a material concern, particularly with the company’s IPO reportedly still on track for November. Enterprise stickiness is a key metric investors will scrutinize. If flagship customers are pulling back, the growth narrative gets harder to sustain at IPO valuations.

Microsoft’s pullback adds a different dimension. Microsoft has its own deep OpenAI relationship and has been integrating Copilot across its product stack. Maintaining a separate, significant Anthropic dependency may simply have been redundant.

Is SpaceX’s AI rebrand driving real value or just noise?

SpaceX stock has climbed to its highest level since June, fueled partly by Starship progress and partly by the AI business narrative. The planned rename of the AI unit to SpaceXSI (Super Intelligence) is the kind of branding move that gets attention, but investors should treat it with appropriate skepticism on its own.

Morgan Stanley analyst Adam Jonas has argued that investors may be undervaluing SpaceX’s long-term growth opportunity, suggesting the stock has become cheaper relative to that opportunity as it pulled back from earlier highs. That framing supports the recent bounce. A 60% recovery off the August low is substantial by any measure.

Motley Fool, however, cautions that most investors are still overpaying at current prices, arguing that the risk profile is being ignored in the enthusiasm around AI and Starship. SpaceX remains a private company accessed through secondary markets, which adds liquidity and price-discovery complexity that public equity investors may underweight.

The AI unit rebrand itself carries limited fundamental weight until Musk provides concrete revenue or customer disclosures for SpaceXSI. Renaming a division is not a business model change.

Nvidia demand keeps building at the infrastructure level

Away from the headline stories, AM Intelligence announced firm orders for 20,000 Nvidia Rubin GPUs, specifically the Vera Rubin NVL72 rack-scale systems, bringing its total planned fleet to 29,000 GPUs. The company, established by the promoters of Indian energy group Greenko, is building AI infrastructure in Hyderabad targeting cloud providers and enterprises.

This is a useful data point on Nvidia’s demand pipeline. The Rubin architecture represents Nvidia’s next-generation GPU platform, and seeing binding orders of this scale from an emerging-market infrastructure player reinforces that demand is not concentrated solely in US hyperscalers. For investors tracking Nvidia’s forward revenue visibility, order announcements like this one contribute to the broader picture, even if AMI is a smaller buyer relative to the Amazons and Microsofts of the world.

A separate analysis on Nvidia’s valuation notes the company is becoming a focal point in the ongoing debate about whether AI investment levels are sustainable. The counterargument to bubble fears, broadly, is that orders like AMI’s reflect genuine infrastructure buildout rather than speculative paper demand. That debate is unlikely to resolve quickly.

Nothing in this article is investment advice. All figures are sourced from the publications cited and have not been independently verified.