Microsoft Anthropic Deal Puts a $43B Azure Bet on Claude

Microsoft Anthropic Azure revenue forecast now has a $43 billion number attached to it. HSBC analyst Stephen Bersey estimates the Anthropic partnership could become that large for Microsoft by 2030 if Claude demand scales fast, Anthropic spends 60% of revenue on compute, and Azure captures 30% of that spend. The public contract already has weight: Anthropic committed to buy $30 billion of Azure capacity and contract up to one gigawatt under the Microsoft, NVIDIA and Anthropic partnership announcement.

The important part for Azure is the shape of the bet. Microsoft is not trying to make Claude exclusive. It is trying to make a fast-growing model company spend across Azure while Amazon and Google fight for the same workloads.

HSBC Turns a Small Backlog Line Into a Big Azure Number

Bersey’s calculation starts with an aggressive Anthropic revenue ramp. The HSBC note pegs Anthropic’s possible 2030 revenue near $241 billion, compared with less than $5 billion this year. If 60% of that revenue goes to infrastructure, the compute wallet would be about $144.6 billion. A 30% Azure share would put the Microsoft piece at roughly $43.4 billion a year.

That number is a forecast, not a contracted sale. The more grounded clue sits in remaining performance obligations (RPO, contracted revenue not yet recognized). HSBC says Anthropic accounts for about 5% of Microsoft’s RPO, while OpenAI accounts for about 46%. In plain English, Claude is still a small line in the backlog, but the line has room to grow if customer use catches up with capacity.

• $43.4 billion – implied annual Azure revenue from HSBC’s Anthropic scenario.
• $30 billion – Anthropic’s public Azure compute purchase commitment.
• $627 billion – Microsoft’s commercial RPO at the end of the March quarter, according to Microsoft’s fiscal third quarter earnings release.
• 40% – Azure and other cloud services revenue growth in the same quarter, or 39% in constant currency.

The gap between 5% and 46% explains why the HSBC note landed. A small backlog contribution does not need to become OpenAI-sized to matter. It only needs to move from side bet to major buyer.

The Compute Stack Around Claude Is Now Multi-Cloud

Anthropic has turned its compute shortage into a buying tour. Microsoft gets Claude on Azure and in Microsoft Foundry. Amazon keeps primary training status. Google and Broadcom sell access to Tensor Processing Units (TPUs, Google-designed artificial intelligence accelerators). SpaceX supplies extra graphics processing unit capacity, with graphics processing units (GPUs, chips suited for parallel AI work) still central to the AI buildout.

The table cuts against a simple winner-take-all read. Claude’s growth helps Azure, but Anthropic is using every credible supplier it can find. That makes HSBC’s 30% Azure share the hinge in the model.

Claude Gives Azure a Second Model Anchor

Microsoft’s OpenAI relationship remains the biggest AI asset in Redmond’s cloud story. The revised agreement says OpenAI products ship first on Azure unless Microsoft cannot and chooses not to support the necessary capabilities. It also makes Microsoft’s license to OpenAI intellectual property non-exclusive, according to Microsoft’s April OpenAI partnership update.

Claude changes the sales pitch. Azure buyers can access Anthropic frontier models through Microsoft Foundry, while Microsoft has committed to continuing Claude access across GitHub Copilot, Microsoft 365 Copilot and Copilot Studio. That matters to chief information officers who do not want one model provider determining their entire AI plan.

• Model choice – Azure can sell OpenAI and Anthropic access to the same enterprise accounts.
• Backlog diversity – a larger Claude footprint would reduce the optics of relying on one partner for AI cloud demand.
• Product pull-through – Claude inside Copilot products gives Microsoft another route from model usage to paid software seats.

For Anthropic, the same structure keeps leverage intact. Claude can sit on AWS Bedrock, Google Cloud Vertex AI and Microsoft Foundry at the same time. For Azure, the win comes from being a large enough supplier in a multi-cloud world.

Capacity, Not Demand, Sets the Ceiling

Microsoft’s latest quarter already shows the bottleneck. Satya Nadella, chairman and chief executive officer of Microsoft, told investors that the company’s AI business passed a $37 billion annual revenue run rate, up 123% year-over-year. The cloud line was strong enough that Intelligent Cloud revenue hit $34.7 billion for the quarter.

Strong customer demand across workloads, customer segments, and geographic regions continues to exceed available capacity.

Amy Hood, executive vice president and chief financial officer of Microsoft, gave that warning on Microsoft’s fiscal third quarter earnings call. She also said capital expenditures (capex, money spent on long-lived and short-lived physical assets) were $31.9 billion in the quarter, with roughly two thirds tied to short-lived assets, mainly GPUs and central processing units (CPUs, general-purpose server processors).

The spending will not ease quickly. Hood said Microsoft expects more than $40 billion of capex in the June quarter and roughly $190 billion for the calendar year. She also said the company expects to remain capacity constrained at least through the end of the year.

That is the hard limit on the HSBC case. Azure can win share only if Microsoft has enough power, chips, cooling and data center space to sell.

Margins Will Decide How Much of the Bet Sticks

A $43 billion revenue opportunity sounds cleaner than it is. AI cloud revenue carries heavy upfront costs, and Microsoft is already showing the strain. Microsoft Cloud gross margin was 66% in the March quarter, down from the prior year because of AI infrastructure investment and rising AI product usage.

The June-quarter guide points to about 64% cloud gross margin. That is still a rich software-company number by most standards, but the direction matters. Training clusters, inference servers, networking gear and power contracts have to be bought before the customer revenue arrives. Some of the gear also ages fast.

The circular nature of these AI deals adds another reason for caution. Microsoft invests in Anthropic. Anthropic buys Azure. Amazon invests in Anthropic. Anthropic buys AWS. NVIDIA invests in Anthropic while selling the systems that make the capacity possible. The revenue is real when compute is consumed, but the strategic money and supplier money are moving through the same small group of companies.

For investors, that makes Bersey’s $571 price target and Buy rating less important than the operating proof. The Microsoft case improves if Azure sells high-use AI capacity at acceptable margins. It weakens if the cloud providers keep bidding up power, land, chips and partner commitments faster than they can bill customers.

The Timeline Runs Through Claude’s Own Numbers

Anthropic gave Microsoft bulls a fresh data point last week. The company said it raised $65 billion in Series H funding at a $965 billion post-money valuation, and said run-rate revenue crossed $47 billion earlier in May, according to Anthropic’s Series H funding announcement. Krishna Rao, chief financial officer of Anthropic, said the capital would help meet demand, fund research and bring Claude to more workplaces.

The startup also said on Monday that it confidentially submitted a draft registration statement for an initial public offering (IPO, a process for selling shares to public investors). No share count or price has been set. Even without public financial statements, the speed of the revenue claims explains why HSBC is willing to run a large 2030 scenario.

There is still a climb from $47 billion of run-rate revenue to $241 billion of annual revenue. That would be a little more than a fivefold increase. Measured against HSBC’s estimate of less than $5 billion for this year, the 2030 figure would be more than 48 times larger. Both comparisons assume the run-rate measure turns into durable billed revenue, not a burst of peak demand.

If Claude’s growth keeps outrunning capacity, Microsoft’s Anthropic deal can become a serious Azure revenue line. If Amazon, Google and new compute suppliers absorb the larger share, the $43 billion number stays where it began, inside an analyst model.

Disclaimer: This article is for informational purposes only and does not provide investment advice. Securities and AI infrastructure spending carry valuation, execution, liquidity and market risks. Consult a qualified financial professional before making investment decisions. Figures are accurate as of publication.