AI has dominated the spotlight over the past year, sparking major changes in the way we work and discuss the future. In line with this trend, Microsoft is going all in on AI, with a particular focus on its Copilot technology.
During his keynote at Microsoft Ignite 2023, CEO Satya Nadella highlighted the rapid advancements in AI since the introduction of ChatGPT a year ago. He said Microsoft and its customers have entered a new era of AI.
“We’re entering this exciting new phase of AI where … we’re getting into the details of product making, deployment, safety, real productivity gains – all the real-world issues,” Nadella said.
Nadella called this phase “the age of Copilots,” pointing to the adoption of Microsoft Copilot by diverse companies, including digital firms like Duolingo, AirBnB, and Shopify, as well as larger entities like BT Group, Bayer, and Lumen. He added that companies across various industries, such as financial services, manufacturing, and healthcare, are building and deploying industry-specific Copilots.
A new Microsoft report on Copilot users showcased several compelling productivity gains. A survey of 297 users in the Copilot for Microsoft 365 Early Access Program found the following:
- 77% of Copilot users said they would not want to give it up.
- 70% said they were more productive.
- 68% credited Copilot with jumpstarting their creative processes.
- 67% said Copilot saved them time so they could focus on more important work.
Additionally, in a study of 6,500 Microsoft customer service and support agents, users saw a 12% reduction in time spent resolving customer issues when using Copilot in Dynamics 365 Customer Service, the report said. A survey of 133 Microsoft salespeople who used Copilot for Sales saved an average of 90 minutes per week.
“The way to think about this is that Copilot will be the new UI that helps us gain access to the world’s knowledge and your organization’s knowledge,” Nadella said. “But most importantly, it’s your agent that helps you act on that knowledge.
Sustainability Commitment for AI Infrastructure
Microsoft is emphasizing it role as a provider of AI infrastructure, particularly through its Azure platform, which Nadella referred to as “the world’s computer.” With more than 60 data center regions worldwide, the company aims to redefine operations across its fleet and data centers in the new age of AI.
Nadella discussed Microsoft’s focus on developing sustainable practices in powering its data center. “Today, we are one of the largest buyers of renewable energy around the globe,” Nadella said, noting that Microsoft has acquired over 19 gigawatts since 2013. He added that Microsoft is committed to working with its partners to diversify its energy sources, incorporating wind, solar, geothermal, and even nuclear fusion.
Additionally, Nadella revealed that the company is on track to meet its goal of generating 100% of the energy used in its data centers from zero-carbon sources by 2025.
Other announcements related to AI infrastructure included:
- Azure Cobalt, the company’s first custom, in-house CPU series, beginning with Cobalt 100, will be available to customers next year.
- Azure Maia, the company’s first custom, in-house Microsoft Cloud AI accelerator optimized for LLM training and inference.
Microsoft CEO Satya Nadella shows the Azure Maia 100 chip
- The general availability of Azure Boost, a system that offloads server virtualization processes traditionally performed by the hypervisor and host OS onto purpose-built hardware and software, such as networking, storage, and host management, according to Microsoft.
- Early access for select customers of AMD’s MI300X Accelerated Azure Virtual Machines for large generative AI workloads.
- The preview of Azure Confidential GPU VMs, an offering that is codesigned with Nvidia.
Microsoft All in on OpenAI and Open Source
In his keynote, Nadella emphasized Microsoft’s deepening partnership with OpenAI. “We are thrilled to be all in on this partnership together,” he said. As part of the partnership, he announced the integration of GPT-4 Turbo and GPT-4 Turbo with Vision into Azure OpenAI Service.
GPT-4 Turbo offers lower pricing, structured JSON formatting, and extended prompt length, allowing users to input substantial amounts of text – up to 300 pages – into a single prompt, Nadella said. The GPT-4 Turbo preview is set to be available in Azure OpenAI Service this week, with token pricing models aligned with OpenAI’s.
Nadella touched on future developments, such as the integration of GPT-4 Turbo with Vision into Azure AI Vision, allowing users interact with video images and text prompts. In addition, Microsoft will introduce fine-tuning capabilities for GPT-4 and Azure OpenAI Service, which will let organizations create custom versions of GPT-4 using their own data.
Nadella also highlighted Microsoft’s commitment to open source, particularly with its inclusion of diverse open source models in Azure. “We want to bring in the best selection of open source models to Azure and do so responsibly,” he said. Microsoft’s model catalog already has an extensive selection, with additions like Stable Diffusion, Code Llama, Mistral 78, and Nvidia Nemotron-3 8B. “All these capabilities are deeply, deeply integrated with our safety guardrails,” he noted.
In that spirit, Nadella unveiled a new Model as a Service offering in Azure. The service aims to provide organizations with access to large open source models without requiring developers to handle GPU provisioning. Microsoft is partnering with Meta on the Model as a Service offering, starting with Llama 2 as a Service. Ultimately, Nadella said Microsoft wants to support models in every language and country.
Other announcements related to foundational models and AI tool chains included:
- Nvidia AI Foundry Service on Azure, an end-to-end offering for creating custom generative AI models.
- Phi-2, the latest version of the Phi small language model built by Microsoft Research.
Microsoft Rolls Out Fabric Data Platform
Nadella discussed Microsoft Fabric, a data platform designed to integrate an organization’s data and analytic workloads into a unified experience. During his keynote presentation, Nadella revealed that Microsoft Fabric has reached general availability.
Additionally, Nadella announced the introduction of Mirroring in Microsoft Fabric. This feature is designed to support the connection between existing data houses and data lakes to OneLake.
Apps Reimagined in the Era of AI
Nadella noted Microsoft’s strategic shift in integrating AI capabilities into its core applications. He focused on Microsoft Teams as an example. Currently, Microsoft Teams’ user base exceeds 320 million people.
“Our vision for Teams has always been to bring together everything you need in one place, across collaboration, chat, meetings, and calling,” Nadella said.
The new version of Teams is up to two times faster and uses 50% fewer resources, according to Nadella. More importantly, he said the revamped Teams will serve as “the foundation for the next generation of AI-powered experiences transforming how we work.”
He also announced that Mesh on Microsoft Teams, a mixed reality platform, will be generally available in January.
‘The Copilot Company’
Nadella emphasized in his keynote that Microsoft envisions itself as “the Copilot company.” “We believe in a future where there will be a Copilot for everyone and everything you do,” he said.
The vision starts with search functionality, now built into Copilot. “Search as we know it is changing, and we’re all in,” Nadella said.
Bing Chat has undergone a shift, now rebranded as Copilot, and operates as a standalone destination. It works across various browsers, including Microsoft Edge, Google Chrome, Safari, and soon-to-come mobile apps. The enterprise version of Bing Chat, which features enhanced commercial data protection, has also been rebranded as Copilot.
Finally, Nadella unveiled Copilot Studio, a platform that lets organizations extend Copilot to their unique needs and systems. Copilot Studio provides tools to build custom GPTs, develop new plugins, orchestrate workflows, monitor Copilot performance, and manage customizations.
Other Copilot announcements included Microsoft Copilot for Service, a Copilot designed for customer service agents.
