Why is Microsoft cutting ties with Nvidia? Billions of dollars at stake

Microsoft is limiting its reliance on Nvidia to avoid billions in infrastructure costs, opting for its own AI solutions for the development of ChatGPT 6. This decision, confirmed in June 2026, drastically changes the balance of power in the technology sector. Satya Nadella's company is moving to a proprietary hardware model, ending the era of uncritical funding for an external computing fleet.

The end of the monopoly era

Tech giants have stopped playing on the same team as Nvidia. The confirmation of this fact on June 15, 2026, marks a turning point in how data centers for next-generation language models are built. Microsoft has withdrawn from its previous model of close cooperation, which guaranteed a smooth supply of hardware for years but simultaneously drained operating budgets. The change in strategy means abandoning Nvidia's exclusivity as the primary supplier of accelerators for large-scale projects.

OpenAI, while developing ChatGPT 6, faced a performance challenge that could no longer be solved by simply purchasing more DGX-type units. Microsoft, as the primary technology partner, assessed that further scaling based on external hardware was economically unjustified. The company stopped treating purchases from the Santa Clara giant as an investment in stability and began viewing them as a financial drag on its own development. This is not a course correction, but a break from a model in which one corporation controls not only the software but also the physical computing layer of the entire ecosystem.

This decision strikes at the foundations of Nvidia's hegemony in the AI market. In recent years, DGX units have been the standard for training large language models. By taking control of the technology stack, Microsoft is aiming for full vertical integration. It wants to manage silicon on its own terms, optimizing processor architecture for the specific operations performed by ChatGPT 6. For the board in Redmond, this means taking responsibility for every stage of the process – from circuit board design to the final results generated by the model.

Microsoft server room prepared for the new AI architecture.

The math of savings

Microsoft has challenged the market hegemon by basing its strategy on a cold analysis of operating costs. AI infrastructure spending has been growing exponentially in recent years. Each subsequent computing cluster based on Nvidia technology involved a massive manufacturer margin, which became an unacceptable budget burden for Microsoft. The decision from June 2026 is a direct response to these rising costs. The company assumes that developing its own chips will allow for a reduction in expenses by billions of dollars annually, which previously flowed to external contractors.

The scale of investment in data centers is so large that even a small optimization in the cost of a single query to the ChatGPT 6 model brings multi-million dollar savings on a quarterly basis. Microsoft no longer wants to fund Nvidia's margins while designing its own hardware. The company is betting on its own silicon, which in theory should allow for better integration of software with the hardware layer. Not having to adapt systems to universal external solutions provides room for significantly increasing energy and computational efficiency.

For investors watching these moves, the math is ruthless. By giving up on purchasing DGX units, Microsoft regains financial liquidity that it could invest in further expansion of cloud services. Optimizing the lifecycle of ChatGPT 6 requires computing power that, at current market prices for GPUs, exceeds profitability assumptions. Microsoft is taking the helm so as not to be a hostage to the pricing policies of hardware suppliers. This is a risky approach because, in the event of a technical failure, the company has nowhere to retreat, but the potential gains from self-reliance are too tempting to pass up.

The architecture of independence

Transitioning to proprietary solutions is open-heart surgery for the largest data centers. Microsoft must prove that its own computing units are capable of bearing the load that DGX Spark systems have handled until now. Standards like OpenClaw, which Nvidia was promoting as recently as March 15, 2026, were supposed to be the foundation for modern clusters. Microsoft, however, decided it wanted to go beyond the imposed framework, creating an architecture that would directly address the specific requirements of the ChatGPT 6 model.

Control over the entire technology stack is not just a matter of savings. It is a matter of operational sovereignty. Dependence on the availability of Nvidia chips put Microsoft in the position of a supplicant. In the event of delivery delays, the schedule for work on the ChatGPT 6 model was at risk. Now, the company is taking responsibility for the supply chain, which in theory should increase resilience to market fluctuations. However, if Microsoft's proprietary chips prove less efficient than Nvidia's solutions, delays in the model's launch will be inevitable, and the reputational damage could be severe.

Analysts point to the technical aspect of this change. Nvidia built its success on the CUDA ecosystem, which became an industry standard. Microsoft, by creating its own chips, must face the necessity of rewriting a significant portion of its software to fully utilize the potential of the new hardware. This is a tedious and risky process, requiring enormous engineering effort. Nevertheless, the decision to move away from external suppliers indicates that the company is ready to bear these costs just to become independent of external technological limitations.

Close-up of a next-generation AI processor in Microsoft labs.

A turning point for ChatGPT 6

The development of ChatGPT 6 has become a catalyst for changes in the relationships between tech giants. This model requires unprecedented computing power, and its role in Microsoft's product portfolio makes it a strategic priority. June 2026 brought a clear message: the development of this service cannot depend on external suppliers. Microsoft is redefining its role, moving away from the "customer buying power" model toward an "infrastructure creator" model.

A Geeky Gadgets report from June 15, 2026, confirms that the situation for OpenAI has become complicated. Changing hardware suppliers in the middle of work on such an advanced model is a process that carries the risk of optimization errors. However, for the board in Redmond, the priority is long-term stability, not just short-term results. Microsoft wants to be sure that in the coming years it will have infrastructure that it fully controls, regardless of competitor moves or changes in Nvidia's sales policy.

This approach also marks the end of the era in which one company could dictate terms to the entire AI sector. Other giants, observing Microsoft's actions, will certainly analyze their own infrastructure costs. If the self-sufficiency model proves effective, the AI hardware market will face drastic reshuffling. Nvidia will cease to be the sole point of reference, and control over technology will become the new battlefield. Microsoft has bet everything on one card – its own independence – and now must prove that its proprietary infrastructure can meet the requirements it set for itself within the ChatGPT 6 project.

Market reaction to the change of front

The stock market landscape for Nvidia after June 2026 looks completely different than it did just a year earlier. Once it became clear that Microsoft was limiting chip orders, the foundations of the graphics chip manufacturer's previous hegemony began to lose stability. The market, accustomed to the constant growth of Nvidia's revenue thanks to the appetite of the largest cloud corporations, had to revise its expectations. Jensen Huang, head of Nvidia, faces the necessity of finding new sources of revenue to fill the gap left by a client as significant as Microsoft.

Investors are watching these shifts with anxiety. Nvidia is losing its status as the exclusive supplier for key AI projects, which in practice means a decline in margins in the data center segment. The company, which until recently dictated terms to the entire sector, is forced to revise its sales strategy. Tightening cooperation with smaller players is becoming a necessity, not a choice. The technology sector, saturated with existing products, reacts nervously to any information about attempts to find new paths of monetization.

Will Nvidia survive without its main client? This question is no longer a theoretical debate, but a real operational challenge. The company must prove that it can sell not only to the biggest players but also to smaller entities that do not have the resources to build their own independent infrastructure. The situation in the AI server market has become more dynamic and less predictable. If Microsoft succeeds with its own solutions, Nvidia will lose not only revenue but also influence over the direction of AI technology development, which until now was largely determined by its products.

Stock market analysts tracking the volatility of tech giant stocks.

The risk of self-sufficiency

The decision made in June 2026 is a strategic turn toward full technological sovereignty, which carries enormous operational risk. Microsoft, by cutting ties with Nvidia, is betting everything on one card. Building one's own chips is not just a matter of design, but also of ensuring production continuity and advanced technical support. If the company's own chips fail at the training stage of the ChatGPT 6 model, the reputational and operational losses could far exceed the billions of dollars saved.

Investors are asking a simple question: is Microsoft capable of maintaining the pace of innovation without the support of the most efficient cards available on the market? For years, Nvidia has become synonymous with AI, offering hardware that defined industry standards. Now Microsoft is challenging this monopoly, risking the stability of its flagship services. The answer to this question is not clear-cut. For years, the company has accustomed us to the high quality of its cloud solutions, but building one's own processors is a completely different league of difficulty.

We will feel the effects of this move in the coming quarters. The entire IT sector is watching this divorce with attention, wondering if other corporations will follow in Microsoft's footsteps. If the self-sufficiency model proves effective, Nvidia will stop dictating pricing terms, and the market for data center hardware suppliers will face drastic reshuffling. In this game, the stake is control over the "brains" of future AI systems. Whether the risk pays off, we will find out when the first computing clusters based on proprietary silicon start actually handling user queries. For now, we are left to observe how giants who were partners until recently stand on opposite sides of the barricade, each with their own plan for dominance in the world of artificial intelligence.

What does this mean for the reader?

From a market perspective, Microsoft is betting on autonomous control over its own destiny, which for Nvidia means a painful lesson about the nature of business dependence. Microsoft shareholders will gain if the company successfully lowers operating costs, but Nvidia investors will lose, as they must reckon with the risk of losing a primary source of revenue. The catch remains the massive technological risk in transitioning to proprietary chips – if performance turns out to be lower than assumed, ChatGPT 6 users may experience a slowdown in service development.

Questions and answers

Why did Microsoft decide to limit cooperation with Nvidia now?

The decision stems from rising infrastructure maintenance costs combined with a desire to gain full control over the development of systems like ChatGPT 6, which will allow for long-term budget optimization.

Will Nvidia lose its position as the AI market leader?

Nvidia remains an innovation leader, however, the loss of its status as the exclusive supplier for Microsoft forces the company to diversify its client base and seek new sources of revenue.

What savings does Microsoft anticipate?

It is estimated that the change in procurement strategy will save billions of dollars annually, which were previously spent on purchasing DGX units and funding the external supplier's margins.

Will Microsoft's own solutions match the performance of Nvidia cards?

This is the biggest unknown of this strategy. Microsoft risks delays in the development of the ChatGPT 6 model if its proprietary chips do not achieve the required technical parameters in performance tests.

Could other tech companies follow in Microsoft's footsteps?

The self-sufficiency model is attractive to the largest cloud players, but it requires enormous capital and engineering resources, which limits this group only to the largest technology corporations.

What will happen to Microsoft's existing computing clusters?

Microsoft will gradually phase out external units in favor of its own infrastructure, which will allow for a smooth migration of services without suddenly interrupting the continuity of work on ChatGPT 6.

Sources

Article prepared by the Wiadomości PRO editorial team with the support of artificial intelligence. Facts are derived from the sources listed above.