Anthropic, Microsoft, and NVIDIA announced a three-part strategic agreement on November 18, 2025: Claude models for Microsoft Foundry, a $30 billion Anthropic commitment to Azure compute plus up to one gigawatt of additional capacity, and joint Anthropic–NVIDIA work on model and hardware optimization. The original preview has since become a production service: Microsoft says Claude in Foundry reached general availability on June 29, 2026.
The arrangement expands enterprise access to Claude without making Claude a Microsoft model or making NVIDIA Anthropic’s exclusive hardware partner. Anthropic continues to identify Amazon as its primary cloud provider and training partner, while also using Google TPUs and NVIDIA GPUs.
What the November 2025 agreement included
| Parties | Commitment | Practical meaning |
|---|---|---|
| Anthropic | Purchase $30 billion of Azure compute capacity and contract for up to one gigawatt of additional capacity | Large-scale Azure infrastructure for training and inference |
| Anthropic and NVIDIA | Joint model, hardware and software design work, initially targeting Grace Blackwell and Vera Rubin systems | Closer alignment between Claude workloads and NVIDIA’s accelerator roadmap |
| NVIDIA | Commit to invest up to $10 billion in Anthropic | Strategic and financial exposure to Anthropic’s growth |
| Microsoft | Commit to invest up to $5 billion in Anthropic | Deeper Azure and Copilot integration |
| Microsoft and Anthropic | Offer Claude through Microsoft Foundry and Microsoft’s Copilot products | Azure customers can select Claude alongside other frontier models |
The investment figures are maximum commitments announced by the companies, not proof that the full amounts had already been transferred on announcement day. Likewise, describing the package as a “$45 billion deal” combines different commitments and can falsely suggest one cash transaction.
Anthropic’s announcement is available at Anthropic’s strategic-partnerships release.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
Claude in Microsoft Foundry: from preview to general availability
Claude Sonnet 4.5, Claude Haiku 4.5 and Claude Opus 4.1 entered Foundry in public preview on November 18, 2025. Microsoft announced general availability on June 29, 2026, so “coming to Foundry” is now outdated framing.
What Foundry provides
- Model deployment through Foundry APIs and Python, TypeScript and C# SDKs.
- Microsoft Entra ID authentication and Azure role-based access control.
- Existing Azure billing arrangements and eligibility for Azure Consumption Commitment, subject to each customer’s contract.
- Foundry Agent Service for agent orchestration, with Microsoft IQ and other enterprise grounding capabilities.
- Global and US data-zone choices in the generally available service.
- Support for capabilities such as tool use, citations, vision, code execution, web search, web fetch, tool streaming and prompt caching, where supported by the selected model and deployment.
Foundry is an enterprise AI development and deployment platform, not merely a model catalogue. Anthropic remains the model provider; Azure supplies the delivery, identity, governance, billing and application environment. Microsoft’s general-availability announcement says Anthropic operates inference and is the data processor and SLA provider, an important distinction for contracts and compliance reviews.
What serverless means
The initial deployment was serverless. Customers call the model without managing the underlying GPUs, but serverless does not mean unlimited capacity or no infrastructure-related constraints. Quotas, rate limits, regional availability, latency, token usage and tool charges still affect an application.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
The current Microsoft description says Foundry runs Claude on NVIDIA Blackwell Ultra systems connected with InfiniBand networking. That is a description of this Foundry service, not a claim about every Claude workload.
Why Azure customers may care
For an Azure-standardized organization, the main benefit may be procurement and control rather than another consumer chatbot. Existing Azure agreements, Entra identities, network policies, governance tools and usage reporting can remain part of the deployment. A company can also evaluate Claude and OpenAI models within one Azure environment and build agents around Claude without moving its broader workflow to a separate cloud.
Anthropic specifically presents Foundry as a way to reduce separate vendor contracts and billing systems. That is a vendor-stated benefit, not a guarantee that every company will eliminate a direct Anthropic relationship or obtain identical commercial terms.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
What “adopting NVIDIA architecture” actually means
The NVIDIA language describes co-design and optimization, not exclusive dependence. Anthropic and NVIDIA plan to optimize Anthropic models for NVIDIA systems and future NVIDIA architectures for Anthropic workloads. The initial capacity commitment referenced Grace Blackwell and Vera Rubin systems; Microsoft’s later Foundry description specifies Blackwell Ultra and InfiniBand.
Potential technical objectives
- Higher inference throughput and better performance per dollar.
- Improved energy efficiency and total cost of ownership.
- Faster scale-out through high-bandwidth networking.
- Closer compatibility between Anthropic’s serving software and NVIDIA’s accelerated-computing stack.
The public announcements do not provide independent latency, tokens-per-second, power-efficiency or cost benchmarks. It is therefore not established that NVIDIA hardware will make every Claude application faster or cheaper. Customers should treat the partnership as an optimization effort, not a performance guarantee.
Anthropic has not abandoned AWS or Google
Anthropic says Amazon remains its primary cloud provider and training partner. It also describes Claude workloads running across AWS Trainium, Google TPUs and NVIDIA GPUs. The Microsoft agreement therefore fits a multi-platform strategy:
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
- AWS: primary cloud and training relationship.
- Google Cloud: TPU capacity and a separate compute partnership.
- NVIDIA: GPU infrastructure and architecture collaboration.
- Microsoft: Azure capacity, Foundry distribution and Copilot access.
Sources: Anthropic’s Google and Broadcom announcement and the strategic-partnerships announcement.
What changes for Microsoft Copilot users
Anthropic said Claude access expanded across GitHub Copilot, Microsoft 365 Copilot and Copilot Studio. Claude powers the Researcher agent in Microsoft 365 Copilot, supports custom agents in Copilot Studio and appeared as an option in Excel’s Agent Mode preview for formula generation, data analysis, error identification and iterative spreadsheet work.
These are product-specific availability statements, not evidence that every Copilot request is automatically routed to Claude. Model selection and feature availability depend on the product and preview or production status.
Best Value
- Powered by the NVIDIA Blackwell architecture and DLSS 4 OC mode: 2640MHz/Default mode: 2610MHz (Boost Clock)
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.125-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
Models and the original Foundry pricing
Microsoft’s November 18, 2025 announcement listed these Global Standard prices per one million tokens:
| Model | Input | Output | Status at launch |
|---|---|---|---|
| Claude Haiku 4.5 | $1 | $5 | Public preview |
| Claude Sonnet 4.5 | $3 | $15 | Public preview |
| Claude Opus 4.1 | $15 | $75 | Public preview |
These were the prices published with the November 2025 announcement, not a complete current Foundry price list for 2026. Newer model SKUs, regions and deployment options can differ. See Microsoft’s original pricing announcement and Anthropic’s pricing reference.
Token rates are only one part of total cost. Tool calls, web search, code execution, storage, networking, logging and agent orchestration can add charges. Global and US data-zone choices may differ in availability or price, while enterprise discounts can change effective rates. Agentic applications can also make many model calls, turning a low per-token price into a substantial bill.
Choosing a Claude access route
| Route | Best fit | Main advantage | Main trade-off |
|---|---|---|---|
| Microsoft Foundry | Azure-centered enterprises | Azure identity, billing, governance, data zones and agent tooling | Greater Azure dependence and possible feature lag |
| Anthropic API | Cloud-neutral developers and teams wanting Anthropic-native features | Direct Anthropic platform relationship | Separate procurement and cloud integration |
| AWS Bedrock | AWS-centered organizations | AWS-native security, networking and data services; Amazon is Anthropic’s primary cloud partner | Less natural for Azure-first estates |
| Google Vertex AI | Google Cloud and TPU users | Google-native data and AI tooling | Less aligned with Microsoft identity and billing |
Official entry points are Microsoft Foundry, the Anthropic API, Amazon Bedrock and Google Vertex AI.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsChecks to complete before deployment
- Confirm that the required Claude version, tools and SDK features are available in the chosen Foundry region and data zone.
- Verify Entra, RBAC, networking, logging and data-processing responsibilities with security and legal teams.
- Check quotas, rate limits, throughput reservations and expected latency; general availability does not mean unlimited capacity.
- Model token, tool and agent-call volume rather than budgeting from input-token prices alone.
- Compare Foundry’s commercial terms with a direct Anthropic contract, Bedrock and Vertex AI, especially if portability matters.
- Run representative workloads before assuming that the NVIDIA collaboration delivers a measurable performance or cost improvement.
Bottom line
This was not simply a model-listing announcement. Anthropic gained Azure distribution and enormous compute capacity; Microsoft gained Claude as a serious enterprise model option and additional Azure and Copilot demand; NVIDIA gained a major model partner and financial exposure to Anthropic. The June 2026 general-availability milestone makes Claude in Foundry a current production route, while Anthropic’s AWS and Google relationships show that the company is diversifying infrastructure rather than moving exclusively to Microsoft or NVIDIA.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
