Overview
On 9 July 2026, Elon Musk posted on X that Anthropic is “obviously currently the leader in AI,” pointing specifically to its Mythos and Fable model families as benchmarks no other company had matched. The statement represents one of the more dramatic public reversals in the recent history of the technology sector. Musk had previously described Anthropic and its Claude models as “woke” and “hypocritical,” and had repeatedly dismissed the company’s prospects in the AI race. His about-face carries weight not simply because of who said it, but because of the commercial arrangement that now sits behind the relationship.
The context is a substantial infrastructure agreement under which Anthropic has committed to paying SpaceX approximately USD 1.25 billion per month through May 2029 for access to computing capacity at SpaceX’s Colossus 1 data centre. That arrangement secures Anthropic access to more than 300 megawatts of power and upwards of 220,000 Nvidia graphics processing units. For enterprise strategists, technology procurement teams, and professional services organisations evaluating their AI vendor settings, the signal here is layered. Musk’s public endorsement of a direct competitor is unusual enough to attract notice on its own terms. Combined with the scale of the underlying commercial deal, it reveals something more fundamental about how the frontier AI sector is now structured.
The development matters for any organisation that has built workflows, procurement decisions, or strategic plans around assumptions about which AI providers are leading, which are stable, and who controls the infrastructure those providers depend upon. Those assumptions are now worth revisiting. The relationship between Anthropic and SpaceX illustrates what industry commentators are describing as the “coopetition” era of advanced AI, a period in which direct competitors must cooperate at the hardware and infrastructure layer to sustain their operations at all.
Key details
The financial scale of the Anthropic-SpaceX infrastructure agreement is significant by any measure. At USD 1.25 billion per month, the annualised commitment approaches USD 15 billion purely for compute access, and the agreement runs through May 2029. The lease covers access to more than 300 megawatts of power capacity and over 220,000 Nvidia GPUs at the Colossus 1 data centre operated by SpaceX. To contextualise that power figure, 300 megawatts is roughly equivalent to the continuous output of a mid-sized gas peaker plant, or approximately the electricity consumption of a city of around 250,000 households depending on usage intensity. The energy footprint of frontier model training and inference is no longer a theoretical concern for corporate sustainability reporting.
Musk’s comments on X specifically cited the Mythos and Fable model families as the benchmark for frontier AI performance as of July 2026. He wrote that “no company has released a model as good as Mythos/Fable” and predicted that Mythos 2 would follow in the near term. Whether Mythos and Fable represent publicly confirmed product designations or internal naming conventions used by Musk was not entirely clear from the available source material at time of publication. What is clear is that the endorsement came from someone with direct commercial exposure to Anthropic’s operational infrastructure and no historical incentive to flatter a rival. Chatham House noted in its July 2026 commentary that the US government had also undertaken a position reversal regarding Anthropic’s Mythos model, though those regulatory events had involved temporary government-mandated rollbacks, adding a further layer of governance complexity to the picture.
The infrastructure dependency this arrangement reveals is structurally important. Anthropic, widely regarded as one of the two or three most capable frontier model developers in the world, is paying a competitor for the physical compute it needs to operate. This is not a minor vendor relationship. At USD 1.25 billion per month, SpaceX is Anthropic’s single largest cost line, and SpaceX’s CEO has now publicly acknowledged that he controls infrastructure Anthropic cannot easily replace. Musk addressed this directly, stating he “would never cut them off in a way that hurt them badly, even as a competitor.” That assurance, however voluntary, highlights the asymmetry in the relationship and the degree to which compute access has become the primary point of advantage in the AI sector rather than algorithmic sophistication alone.
The competitive dynamics this creates extend well beyond the two companies directly involved. When a company of Anthropic’s standing requires this volume of external compute at this cost, it signals that the capital requirements for remaining at the frontier of AI development have moved beyond the reach of all but a handful of organisations globally. For enterprise buyers evaluating which AI providers to build workflows around, the financial and infrastructure resilience of those providers is now a procurement risk category in its own right, sitting alongside considerations of model capability, data governance, and pricing.
Australian context and implications for enterprise AI strategy
For Australian professional services firms, government agencies, and enterprise technology teams, this development lands in a specific context. Australian organisations have been relatively active in adopting large language model tools for tasks ranging from document drafting and data analysis through to regulatory interpretation and client-facing reporting. Many of those deployments rely on API access to models operated by Anthropic, OpenAI, or similar frontier providers. The Anthropic-SpaceX arrangement illustrates that the stability and continuity of those API services is now materially dependent on infrastructure agreements negotiated between US-domiciled private companies operating outside Australian regulatory reach.
References and related sources
- Primary source: www.businessinsider.com
- hyper.ai
- chathamhouse.org
How iEnvi can help
iEnvi integrates technology and data-driven approaches into environmental consulting. We monitor AI and technology developments that affect how environmental professionals deliver services to clients.
This is an iEnvi Machete news summary. Prepared by iEnvi to summarise the source article for environmental professionals tracking AI, data, and technology developments that affect consulting and project delivery.
Published: 10 Jul 2026
Need advice on this topic? Speak to an iEnvi expert at info@ienvi.com.au or 1300 043 684, or contact us online.
Need advice on this issue? iEnvi provides practical, senior-led environmental consulting across contaminated land, remediation, ecology and environmental risk.
Contaminated land advice Remediation services Discuss your site Talk to iEnvi