Anthropic, the AI safety startup behind Claude, is negotiating a second multibillion-dollar infrastructure deal with Nscale, a cloud computing operator, according to a source familiar with the discussions. The two companies previously completed a comparable agreement to develop a data center at the same U.S. location, marking an expansion of their existing partnership.
The talks underscore the massive capital requirements facing AI developers as they race to secure compute capacity for training large language models. Anthropic has emerged as one of the industry's top players alongside OpenAI and Google, competing fiercely for the GPUs and infrastructure necessary to scale operations. Each major AI lab now requires tens of billions of dollars in capital expenditure to maintain competitiveness.
Nscale operates specialized cloud infrastructure built for computationally intensive workloads. The company has positioned itself as a vendor focused on supporting hyperscalers and AI companies with customized data center solutions. By negotiating multiple agreements with a single customer like Anthropic, Nscale secures long-term revenue visibility while gaining leverage with hardware suppliers and energy providers.
Multibillion-dollar compute deals have become the new market standard for frontier AI development. In 2024, Amazon announced plans to invest up to $100 billion in Anthropic as part of a broader commercial partnership. Microsoft similarly committed massive sums to OpenAI for infrastructure and exclusivity arrangements. These arrangements reflect the winner-take-most dynamics of AI: companies must build or contract sufficient compute to train increasingly capable models, or risk falling behind competitors.
Anthropic raised $5 billion in Series C funding in September 2024 at a $20 billion valuation. The round included investments from Google, Salesforce Ventures, and others. The startup competes directly with OpenAI's ChatGPT and Google's Gemini for enterprise adoption and consumer market share. Each generation of model requires exponentially more computation, driving constant capital recycling into infrastructure.
Data center real estate and power availability have become bottlenecks for AI development. Companies now compete not just on model architecture and talent but on securing adequate electricity supply and physical space for servers. This explains the strategic importance of partnerships with infrastructure operators like Nscale. By locking in capacity through multiyear deals, AI companies hedge against supply constraints and price volatility.
The second deal with Nscale suggests Anthropic expects sustained demand for Claude across multiple use cases and geographies. Building redundancy into infrastructure also protects against outages that could disrupt customer operations. As enterprises integrate AI into mission-critical workflows, uptime and geographic distribution become competitive advantages.
Anthropic's infrastructure strategy extends beyond compute procurement. The company has also negotiated cloud credits from Google and maintains relationships with AWS. This diversification reduces dependence on any single vendor, similar to how major tech companies hedge their cloud infrastructure exposure.
The timing matters as well. Regulatory scrutiny around AI intensifies while the incoming Trump administration signals a lighter regulatory touch. Anthropic may be accelerating infrastructure commitments now to lock in expansion plans before potential changes to energy policy or semiconductor export controls affect capacity availability.
Investors tracking Anthropic's trajectory should monitor announcements around compute capacity utilization, customer wins at major enterprises, and whether competing AI labs announce similar or larger infrastructure deals.