This article is part of our series that explores the business of artificial intelligence
We’re seeing the latest chapter of the Microsoft-OpenAI saga unfold after the two companies declared the extension of their partnership. The multi-year, multibillion-dollar investment will provide OpenAI with a lifeline and infrastructure to continue its extremely expensive artificial intelligence research. And Microsoft will have exclusive access to OpenAI’s technology and its top engineering and scientific AI talent.
Both the research lab and tech giant stand to gain from the partnership in the short term. But in the long term, OpenAI will inch toward becoming a Microsoft subsidiary, making it hard for the AI lab to pursue its independent research.
Shared vision, different tone
Both companies speak of a shared vision for AI. In its blog post, OpenAI describes its mission as “to ensure advanced AI benefits all of humanity” and says that it seeks to fulfill its mission without sacrificing its “core beliefs about broadly sharing benefits and the need to prioritize safety.”
“Microsoft shares this vision and our values, and our partnership is instrumental to our progress,” OpenAI’s blog post reads.
Microsoft also speaks of a “shared commitment to building AI systems and products that are trustworthy and safe.”
According to Microsoft CEO Satya Nadella, “We formed our partnership with OpenAI around a shared ambition to responsibly advance cutting-edge AI research and democratize AI as a new technology platform.”
However, there are some interesting differences in the statements of OpenAI and Microsoft. OpenAI describes the investment as “extending our partnership.” On the other hand, Microsoft calls it the “third phase of our long-term partnership.” Microsoft’s description seems to hint that the partnership will evolve into becoming something deeper and more consequential.
OpenAI says that the extension “will allow us to continue our independent research.” Microsoft puts more focus on the product side and says the partnership “enables each of us to independently commercialize the resulting advanced AI technologies.”
I’d doubt that OpenAI could independently commercialize its technology while it is deeply reliant on Microsoft’s cloud computation infrastructure. Microsoft explicitly says that Azure is “OpenAI’s exclusive cloud provider” and will “power all OpenAI workloads across research, products and API services.”
In its blog post, OpenAI says, “Azure’s unique architecture design has been crucial in delivering best-in-class performance and scale for our AI training and inference workloads.” And Microsoft says that it will increase its “investments in the development and deployment of specialized supercomputing systems to accelerate OpenAI’s groundbreaking independent AI research.”
The companies are co-evolving. With the help of OpenAI engineers, the Azure team is developing hardware that is optimized for training and running large language models like GPT-3, ChatGPT, Codex, and DALL-E.
These models are tricky and expensive to run. Getting the hardware right will help OpenAI reduce the costs of research and training but make it more difficult to switch to a different cloud in the future. For Microsoft, better hardware will give it an edge by enabling it to provide AI technology at competitive prices. But Microsoft will also be able to use its infrastructure to deliver AI products that are not related to OpenAI’s technology.
Another interesting detail is OpenAI’s dependence on Microsoft’s market. “Learning from real-world use—and incorporating those lessons—is a critical part of developing powerful AI systems that are safe and useful,” the OpenAI blog reads.
While OpenAI is gathering real-world usage data through its OpenAI API service, it is also largely dependent on Microsoft’s huge cloud market. I have predicted that the Azure OpenAI Service will eat into the market of OpenAI API as it provides a much more reliable platform to build AI products.
At the same time, OpenAI also gets feedback through the integration of its technology into other Microsoft products, including GitHub Copilot, Microsoft Bing, Designer, and Power BI.
Interestingly, Microsoft’s statement strikes a different tone on integrating OpenAI technology into its products, and there is no mention of learning. “Microsoft will deploy OpenAI’s models across our consumer and enterprise products and introduce new categories of digital experiences built on OpenAI’s technology,” the statement reads, indicating that Microsoft is more focused on solidifying its position in current markets and expanding its reach to new ones.
Microsoft acquisition strategy
There is still no talk of an acquisition, and there are several sound reasons for Microsoft not to look for one yet. For one thing, the AI technology that OpenAI develops is still nascent. We are still discovering and exploring its ethical, legal, and commercial challenges. Keeping its distance allows Microsoft to avoid the complications of rolling out new technologies.
Meanwhile, Microsoft has put itself in a very favorable position to acquire OpenAI in the future. Already, the AI lab has become so dependent on Microsoft’s technology and financial backing that it switching to a new strategic partner would be very costly.
And an acquisition in the future seems to be inevitable. The Microsoft blog says that OpenAI’s technology running on Azure has introduced “large-scale AI as a powerful, general-purpose technology platform that we believe will create transformative impact at the magnitude of the personal computer, the internet, mobile devices and the cloud.” Microsoft missed on mobile. I don’t think it wants to miss out on AI.
As I have observed before, when it comes to acquisition, Microsoft is more calculated and less aggressive than other big tech companies.
Microsoft likes to cast a wide net, simultaneously developing ties with many companies in innovative sectors. It usually starts by providing them with subsidized access to Azure. It also makes many of its investments in Azure credits, making sure that companies it invests in will be locked into its platform. This puts Microsoft in a position to both help those companies grow and learn from them. And the investment pays off when the company’s technology and business model mature, bringing in more clients for Azure.
As these sectors mature, Microsoft gradually enters partnerships with the more successful startups. And when the time is right, it acquires companies that will give it the best leverage in the market. Microsoft already did this with Nuance, a company that specialized in AI for healthcare. OpenAI seems to be moving down the same path. When Microsoft closes in for the acquisition, OpenAI will have little choice but to accept.