Microsoft, NVIDIA and Anthropic deepen ties with multibillion-dollar commitments
A new three-way alliance was announced as Microsoft and NVIDIA committed up to $15 billion to Anthropic. The AI firm will in turn purchase $30 billion worth Azure compute and one gigawatt of NVIDIA capacity to advance model performance.
Microsoft, NVIDIA and Anthropic have announced strategic partnerships, with the two investing firms pledging up to $15 billion to the AI developer and forming a three-way collaboration aimed at accelerating the development and adoption of artificial intelligence (AI) across the enterprise sector.
In addition, Anthropic has committed to purchasing $30 billion worth of Azure compute and securing up to one gigawatt of NVIDIA capacity using Grace Blackwell and Vera Rubin systems to support model co-optimisation.
Microsoft Chairman and CEO Satya Nadella framed the initiative around the partners’ complementary strengths, describing NVIDIA as operating at the “silicon layer” and Anthropic at the “cognition layer”. The goal is to shape what developers and organisations can build in the future.
The expanded Microsoft–Anthropic partnership is multifaceted, centring on infrastructure, model access and market reach. The relationship is designed for mutual benefit, with each company increasingly becoming a customer of the other.
Microsoft will integrate Anthropic’s models across its Copilot family, including GitHub Copilot, Microsoft 365 Copilot and Copilot Studio, while Anthropic will scale its rapidly growing Claude model on Azure.
Anthropic co-founder and CEO Dario Amodei noted that this collaboration will make Claude the first frontier model available across all three major cloud platforms. For Microsoft’s enterprise customers, the partnership broadens model choice and provides access to Claude-specific capabilities. Users of Microsoft Foundry will be able to access Anthropic’s frontier models, including Claude Sonnet 4.5, Claude Opus 4.1 and Claude Haiku 4.5.
Both companies plan to go to market together to help customers realise the value of AI, with Amodei citing Microsoft’s strong enterprise reputation. This commercial agreement is backed by Microsoft’s commitment to invest up to $5 billion in Anthropic.
In parallel, NVIDIA and Anthropic are establishing a deep technology partnership to support Anthropic’s future growth.
NVIDIA founder and CEO Jensen Huang expressed enthusiasm about accelerating Claude and praised Anthropic’s work in AI safety, Claude code and the Model Context Protocol, which he said has “completely revolutionised the agentic AI landscape”.
This collaboration centres on hardware optimisation. Anthropic will adopt NVIDIA’s architecture and make extensive use of its accelerators, with joint engineering efforts to co-optimise Anthropic’s models for performance, efficiency and total cost of ownership. Optimisation will begin with the Blackwell architecture and later expand to Vera Rubin systems. The aim is an “order of magnitude speed-up” for Claude, which Huang said will reduce token costs and support broader AI adoption.
Anthropic’s compute commitment includes up to one gigawatt of capacity using Grace Blackwell and Vera Rubin systems. NVIDIA is also backing the partnership with an investment of up to $10 billion in Anthropic.
These moves reflect the wider trend of hyperscalers and model developers forming deeper alliances as competition in frontier AI intensifies. Rapid growth in model size, training costs and inference demand is driving companies to secure long-term access to advanced compute, specialised hardware and cloud infrastructure.
Edited by Jyoti Narayan