A new AI market signal is forming around who can afford the next layer of AI infrastructure.
On July 2, CNBC reported that Nvidia plans to offer startup customers access to revenue-sharing deals. A few days later, CNBC reported that Microsoft was cutting 2.1% of its workforce while its Xbox unit prepared studio changes.
Taken together, the signal is not simply about chips or layoffs. It is about AI infrastructure moving into a more difficult financial stage. Large companies are still spending heavily on AI, but the structure around that spending is changing. Nvidia is looking for ways to help smaller AI cloud companies access expensive compute. Microsoft is reducing headcount and reshaping parts of the business while continuing to absorb higher AI infrastructure costs.
The Public Record:
On July 1, Nvidia published a post titled “NVIDIA Unlocks AI Compute at Scale, Inviting Partners to Power the AI Infrastructure Buildout.”
In that post, Nvidia said it is partnering with AI clouds to deploy large-scale, multi-tenant AI factories through a “revenue-sharing and credit-support model.” Nvidia said the model allows AI cloud companies to sell Nvidia-powered cloud services, while Nvidia earns both standard product revenue and a share of cloud revenue on supported capacity.
Nvidia named Sharon AI and Firmus as early companies working with it under the model.
Nvidia said Sharon AI is deploying up to 40,000 Nvidia Grace Blackwell GB300 GPUs. Sharon AI said its six-year collaboration with Nvidia will enable 72 megawatts of new data center capacity in Australia.
Nvidia also said Firmus is building a DSX AI factory campus in Batam, Indonesia, expected to scale to 360 megawatts and up to 170,000 Nvidia GPUs. Light Reading reported that the Firmus facility is expected to be operational by the first quarter of 2027 and that the partnership includes revenue sharing and credit support.
Separately, Reuters reported on July 6 that Microsoft would cut 4,800 jobs, or about 2.1% of its workforce. Reuters also reported that Microsoft’s Xbox restructuring would involve 3,200 job cuts and divestment of up to five studios.
Microsoft’s own FY26 third-quarter performance report shows why AI infrastructure belongs in the same watch file. Microsoft said gross margin percentage decreased because of continued investments in AI infrastructure and growing AI product usage. It also said operating expenses increased partly because of investments in research and development compute capacity, AI talent, and data.
What the Companies Say:
Nvidia describes the new model as a way to open access to accelerated computing for startups, model builders, enterprises, research organizations, and regional AI players.
The company says emerging AI companies have historically had limited access to capital-intensive infrastructure, even when they had long-term customer commitments. Nvidia’s public post frames the revenue-sharing and credit-support model as a way for AI clouds to procure Nvidia infrastructure and serve AI-native, enterprise, and software customers.
Sharon AI describes itself as an Australian AI cloud and high-performance computing company focused on artificial intelligence and cloud GPU/CPU compute infrastructure. The company says its platform and compute infrastructure are intended to support AI factories and sovereign AI solutions.
Firmus describes its Batam project as an AI factory campus serving AI-native customers. Light Reading reported that Firmus and Singapore-based DayOne are constructing the facility in Indonesia.
Microsoft’s public investor materials describe AI infrastructure as a cost factor across its cloud and productivity businesses. Microsoft also said in its FY26 third-quarter materials that Microsoft 365 Commercial cloud revenue grew with help from Microsoft 365 Copilot, while cost of revenue increased because of AI infrastructure needed to support Copilot seat and usage growth.
Why It May Matter:
This may indicate that AI infrastructure is becoming harder to finance through ordinary hardware sales and cloud contracts alone.
Nvidia’s revenue-sharing model suggests that smaller AI cloud companies may need new financing structures to access large blocks of GPUs. Instead of selling hardware once, Nvidia can participate in cloud revenue tied to supported capacity.
That is a different kind of infrastructure signal. It points to AI compute becoming more like a long-term capacity business, where the question is not only who buys chips, but who can finance, power, fill, and monetize them over time.
Microsoft’s cuts add a second signal. Large AI buyers are still spending on AI infrastructure, but they are also being pushed to show operating discipline. Microsoft’s own investor materials show that AI infrastructure and product usage are affecting gross margin. Reuters reported that Microsoft said the roles eliminated were not being replaced by AI, while also noting Microsoft’s statement that AI is changing how work gets done.
AI infrastructure demand remains strong, but the financial pressure around that infrastructure is becoming more visible.
Related Signals:
-
Nvidia reported record first-quarter fiscal 2027 revenue of $81.6 billion and record data center revenue of $75.2 billion. Nvidia also said it is shifting to a new reporting framework that separates Data Center and Edge Computing, with Data Center including hyperscale and AI-cloud-related categories.
-
Sharon AI said its Nvidia collaboration expands its total AI factory capacity to 132 megawatts, with 102 megawatts contracted to end customers. The company also said it expects to have more than 55,000 total Nvidia GPUs deployed by mid-2027.
-
Data Center Dynamics reported that Sharon AI’s Nvidia agreement uses a revenue-sharing and credit-support structure, with Nvidia earning standard product revenue and a share of cloud revenue on supported capacity.
-
Light Reading reported that Firmus’s Batam campus is designed as a multi-tenant AI-native project, not only a hyperscaler project. It also reported that the agreement covers up to 170,000 Nvidia AI accelerators across future Nvidia platforms through 2027 and 2028.
-
Microsoft said in its FY26 third-quarter investor materials that Microsoft Cloud gross margin percentage decreased because of AI infrastructure investment and growing AI product usage, partly offset by efficiency gains.
What Is Not Known:
The public record does not show the exact revenue-sharing percentages in Nvidia’s arrangements with Sharon AI or Firmus.
It is not known how much customer demand for the planned GPU capacity is firm, cancellable, prepaid, or dependent on future conditions.
It is not known how much of the planned compute capacity will be used for training, inference, fine-tuning, agent workloads, research, enterprise software, or general cloud workloads or whether revenue-sharing models will become common across AI cloud providers or remain limited to selected partners.
It is not known how much Nvidia financial support, credit support, or commercial flexibility is attached to each arrangement beyond what the companies have publicly disclosed.
It is not known how Microsoft’s latest job cuts will affect specific AI teams, cloud operations, gaming operations, or future product lines or whether AI infrastructure costs will continue to pressure Microsoft’s margins at the same rate shown in the latest quarter.
What to Watch Next:
Search terms worth tracking include “Nvidia revenue-sharing model,” “credit-support model AI cloud,” “AI factory financing,” “GPU-backed financing,” “neocloud revenue share,” “Sharon AI Nvidia,” “Firmus Nvidia Batam,” “DSX AI factory,” “Microsoft AI infrastructure margin,” “Microsoft Copilot infrastructure cost,” and “AI cloud capacity agreements.”
Companies worth watching include Nvidia, Microsoft, Sharon AI, Firmus Technologies, DayOne, CoreWeave, Nebius, Crusoe, Nscale, Fluidstack, VAST Data, and major cloud providers building or renting AI capacity.
Infrastructure topics worth tracking include GPU financing, data-center power access, committed cloud offtake agreements, AI cloud utilization, cooling, grid interconnection, sovereign AI clouds, and multi-tenant AI factories.
Source Notes:
CNBC’s July 2 Nvidia report was used as the article’s starting signal for Nvidia’s planned startup customer revenue-sharing access.
CNBC’s July 6 Microsoft report was used as the article’s starting signal for Microsoft’s 2.1% workforce reduction and Xbox restructuring.
Nvidia’s July 1 company post verifies the revenue-sharing and credit-support model, the use of AI clouds, the Sharon AI and Firmus examples, and Nvidia’s description of the compute-access problem for emerging AI companies.
Sharon AI’s June 12 announcement verifies the six-year collaboration with Nvidia, the 72-megawatt Australian capacity plan, the up-to-40,000 GB300 GPU figure, and the company’s description of its AI cloud business.
Data Center Dynamics verifies Sharon AI’s 72-megawatt Nvidia agreement, revenue-sharing structure, 132-megawatt total AI factory capacity figure, and the company’s expected GPU deployment by mid-2027.
Light Reading verifies the Firmus and DayOne Batam project, the 360-megawatt scale, the up-to-170,000 Nvidia accelerator figure, and the reported revenue-sharing and credit-support structure.
Reuters verifies Microsoft’s July 6 job-cut announcement, the 4,800-job figure, the 2.1% workforce figure, the Xbox restructuring, and Microsoft’s statement that the eliminated roles were not being replaced by AI.
Microsoft’s FY26 Q3 performance report verifies Microsoft’s statement that gross margin percentage decreased because of continued AI infrastructure investment and growing AI product usage.
Microsoft’s FY26 Q3 Productivity and Business Processes report verifies that Microsoft 365 Copilot usage contributed to Microsoft 365 Commercial growth and that AI infrastructure increased cost of revenue.
Nvidia’s fiscal Q1 2027 earnings release verifies Nvidia’s $81.6 billion quarterly revenue, $75.2 billion data center revenue, and new reporting framework for Data Center and Edge Computing.
Disclaimer:
This article is for informational and research purposes only. It is based on public records, company materials, and other cited sources available at the time of writing. AICHatterNews.com does not provide investment, legal, business, or financial advice. No statement in this article should be interpreted as a recommendation to buy, sell, invest, or take action involving any company, security, product, or project mentioned. Readers should verify information independently.