Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More
Microsoft made several big announcements about generative AI at its annual Ignite 2023 conference in Seattle. One of the key takeaways was their renewed commitment to open-source generative AI, despite their successful partnership with the closed-source AI leader, OpenAI.
Based in Redmond, Washington, Microsoft has positioned itself as a leader in generative AI by partnering early with OpenAI, integrating OpenAI’s GPT and DALL-E models into its products like Bing Chat, Copilot, and Microsoft’s Image Creator.
But at this year’s Ignite, Microsoft also took significant steps to support open-source AI models from companies like Meta Platforms, which leads in open-source generative AI with its Llama models. These open-source models are now available as services on Microsoft’s Azure cloud platform. Enterprises can fine-tune and deploy both Llama and Mistral 7B—another open-source model—directly on Azure. This move was praised by Meta’s AI pioneer, Yann LeCun.
The implication of providing open-source models on Azure is noteworthy. Since open-source models are free to use (depending on licensing), Microsoft’s decision to support them on Azure might compete with its own investment in OpenAI. Businesses looking for cost-effective AI solutions might opt for these nearly equivalent open-source options over the pricier OpenAI models.
Additionally, Microsoft introduced its new AI model, Phi-2, an upgrade from Phi-1.5. With 2.7 billion parameters, it is designed to operate efficiently on systems with limited GPU power, addressing ongoing GPU shortages. However, Phi-2 is currently available only for research purposes. Sebastien Bubeck, Microsoft’s senior principal research manager, hinted that this might change based on usage and demand.
Microsoft’s support for open-source AI at Ignite 2023 is notable, especially considering CEO Satya Nadella’s recent appearance at OpenAI’s first developer conference, DevDay. This move seems to be part of a broader strategy to offer a variety of AI models to customers, from open-source to high-performance closed-source options, similar to what other cloud providers like Amazon Web Services are doing.
This pivot toward open-source AI could be seen as a shift from Microsoft’s previous exclusive focus on OpenAI. While it may not indicate a rift, it suggests that Microsoft is broadening its AI strategy. By offering both open and closed-source AI models, they are positioning themselves to meet diverse customer needs and stay competitive in the evolving AI landscape.