“There are people inside both companies that hate this thing” — Microsoft and OpenAI remain partners, but for how long?

by

OpenAI, the company behind ChatGPT, and Microsoft have a complex and increasingly challenging partnership.

Now that OpenAI can get computing power from sources other than Microsoft – after their previous exclusive deal ended – the relationship between the two companies is becoming strained and is starting to show signs of trouble.

A recent report from The Information reveals that the partnership between OpenAI and Microsoft isn’t universally popular within either company. According to someone involved in the negotiations, many employees at both OpenAI and Microsoft have concerns about the deal. These concerns seem to be more common among those who aren’t closely involved in the specifics of the partnership.

Even though there’s friction between the two companies, they’re managing to collaborate—at least for the time being. One source explained that while discussions can be difficult, they’re still finding ways to resolve issues and move forward.

Sam Altman, the CEO of OpenAI, has repeatedly asked Microsoft for more computing resources, and Microsoft has been working to deliver as they expand their data operations.

OpenAI’s pursuit of advanced artificial intelligence (AGI) is driving them to seek computing resources from various sources, even though they previously had an exclusive agreement with Microsoft. Their needs appear to be constantly growing.

This isn’t automatically negative. Microsoft’s CFO, Amy Hood, has expressed worry that constantly meeting OpenAI’s growing and costly requests could be detrimental to Microsoft if the AI servers they create don’t become profitable.

OpenAI doesn’t expect to turn a profit until 2029, and is currently losing billions of dollars each year. They now estimate it will cost $115 billion to reach profitability, significantly higher than their initial estimate of $35 billion.

There are people inside both companies that hate this thing.

Recent data from TechCrunch suggests that ChatGPT’s mobile app downloads have leveled off, currently standing at approximately 72 million users after a period of growth.

We expect global downloads to fall by around 8.1% in October. This continues a recent trend, as downloads already decreased significantly from August to September.

While growing competition from other AI is a factor, the lackluster release of GPT-5 in August – a model once predicted to achieve artificial general intelligence – significantly dampened enthusiasm among many users.

Microsoft’s partial breakup with OpenAI might not be such a bad thing.

Microsoft and OpenAI’s relationship — broken promises and billions of dollars

Around six years ago, an AI company was seeking significant investment to develop true artificial intelligence (AGI). Microsoft provided $1 billion, becoming the company’s exclusive cloud provider and gaining access to ChatGPT, which Microsoft then used to create the AI assistant in Windows 11, Copilot.

Microsoft has significantly invested in OpenAI, first with a $2 billion contribution in 2021, followed by a substantial $10 billion investment in 2023. In return, Microsoft is entitled to a portion of OpenAI’s profits and will be OpenAI’s exclusive cloud computing provider through 2030.

Initially, the idea was different. However, in June 2024, OpenAI, Oracle, and Microsoft agreed to a new partnership. This deal uses Oracle’s cloud technology to help Microsoft’s Azure servers run ChatGPT.

Shortly after, OpenAI requested that Microsoft secure another 4.5 gigawatts of power from Oracle, scheduled for delivery in December 2024.

At roughly the same time, concerns started surfacing about limitations in computing power within the AI industry. Leading AI companies began to blame these limitations as the reason they hadn’t yet achieved artificial general intelligence (AGI) – essentially, their ultimate goal.

I’ve been watching the relationship between Microsoft and OpenAI shift recently. It seems they’re becoming more independent – Microsoft cancelled plans for two data center projects they had with OpenAI. At the same time, OpenAI announced a huge undertaking called ‘Stargate’ – a plan to build massive data centers all over the US, and they’re partnering with Oracle and Softbank to make it happen. It’s a $500 billion project, so it’s a significant move.

This change meant Microsoft was no longer the sole cloud provider, although they still have priority for future deals and receive a portion of the earnings.

The disagreement between Microsoft and OpenAI continued as OpenAI secured a $22.4 billion agreement with Coreweave in March 2025, followed by a server rental deal with Google in June.

Last September, NVIDIA pledged $100 billion to OpenAI to accelerate the building of new data centers, aiming for at least 10 gigawatts of power capacity. Shortly after, AMD also announced a collaboration with OpenAI to provide at least another 6 gigawatts of data center power, representing an investment worth tens of billions of dollars.

Everyone’s worried that huge investments from a few AI companies and investment firms could be creating a dangerous bubble. If this bubble bursts, it could cause a market crash unlike anything we’ve seen before.

Microsoft is working to reduce its reliance on NVIDIA and AMD for the powerful chips needed to run artificial intelligence. According to Kevin Scott, Microsoft’s Chief Technology Officer, the company aims to start using its own internally developed chips for AI processing.

Microsoft is reportedly developing its own AI models, though they’re currently a few months behind the cutting-edge technology from OpenAI. According to Microsoft’s AI chief, Mustafa Suleyman, this is a way to reduce their reliance on ChatGPT, as OpenAI hasn’t always provided Microsoft with timely access to its newest AI advancements as agreed.

Read More

2025-10-21 18:43