Just like Google did last week, Microsoft also made a series of major announcements about AI at its annual Build developer conference. To help everyone better understand and digest this news, this article will summarize the key takeaways that I believe are most important. First, I want to express my opinion:

Microsoft provides the ideal platform for developers, especially enterprise developers, who want to build AI products and intelligent solutions through Copilot.

Now, we have witnessed the impressive appearances of these two tech giants, Google I/O and Microsoft Build, and we have a fairly clear understanding of their plans for the next year. But it has to be said that Microsoft's current position is indeed enviable. Their main competitors are Google (with powerful models and large-scale infrastructure) and Apple (with numerous consumer devices and operating systems). However, these two companies have not been able to provide a complete set of AI-era solutions like Microsoft has. Microsoft has demonstrated its strong capabilities in these four areas and more.

Let's start with Google. They have released a series of impressive new AI technologies, including LearnLM, Gemma and Gemini 1.5, as well as some new features for Google Workspace, Google Cloud, and Pixel devices, which are truly powerful. However, they lack a truly powerful PC competitor (Chromebook is obviously not competent), which has become a clear weakness in their AI ecosystem. In addition, the potential transformation of their search business model and the chain reaction it may have on the creator economy are also issues they need to pay attention to in the short term.

Talking about Apple. Although we still need to wait until next month's WWDC (Worldwide Developers Conference) to see what new content they will bring, their threshold is indeed quite high. It needs to be clarified that this is not to belittle Apple, I just think they are not fully prepared yet. I believe they will have a series of exciting announcements and incredible new device releases later this year. If the recent iPad event can be taken as a reference, this year will be their most impressive year in terms of hardware performance.

Apple's M-series custom chips have significantly surpassed their competitors since their release, and the new M4 series also seems to be a heavyweight product in the market. In addition, as a leader in consumer electronics, Apple naturally has no problem with hardware. At WWDC, we may finally see Siri transform into a true AI assistant, a locally running "Apple AI" language model, and a range of new MacOS and iOS AI frameworks.

However, Apple also faces two major problems. First, they lack the key elements needed to build a vertically integrated AI platform, such as their own search engine, large-scale cloud computing infrastructure, and most importantly, their own cutting-edge cloud-based models. These may be gradually built by them in the future, but for now, they have to rely on partners.

Second, data is the key to making AI models truly useful. But Apple's long-standing privacy policy may conflict with the data access permissions required for developers to build useful solutions.

In contrast, Microsoft has every link in the entire technology stack - devices, operating systems, software, development tools, models, and large-scale cloud computing. Therefore, they do not have the above limitations or concerns. In fact, Microsoft has made a lot of investments in talent, technology, and strategy over the past decade to drive the realization of its AI vision. Now, these investments give them enough confidence to prove why the Copilot ecosystem should be the preferred choice for building AI-driven products and experiences.

Here are five important announcements that Microsoft made at this week's Build conference, which further support the previous points:

Copilot + Personal Computers

Microsoft has launched Copilot + Personal Computers, a brand new device category that can be considered the fastest and most AI-ready personal computer built to date. These personal computers are equipped with cutting-edge neural processing units (NPUs), providing unprecedented performance and efficiency for AI workloads with up to 40 TOPS of computing power. The Qualcomm Windows Snapdragon Development Kit, equipped with the Snapdragon X Elite SoC, allows developers to create unique AI experiences that run at the edge. Microsoft even compared it to the MacBook Air M3, showing that the performance of Copilot + Personal Computers far exceeds the latter.

Windows Copilot Runtime

Windows Copilot Runtime is a fundamental change to the Windows operating system aimed at helping developers accelerate AI development on Windows. It includes the Windows Copilot library, a set of APIs supported by AI models on more than 40 devices, as well as AI frameworks and toolchains for introducing custom models into Windows. This end-to-end platform simplifies the complexity of rapidly evolving AI technology, providing developers with a stable target to build intelligent applications.

AI Models

Windows has officially become the first platform to feature the advanced small language model Phi Silica, which is customized for the NPU in Copilot + Personal Computers. Microsoft also announced that OpenAI's flagship multimodal model GPT-4o is now available in Azure AI Studio. GPT-4o supports text, audio, image, and video input/output, enabling developers to transform any application or website into a multimodal, conversational interactive interface. In addition, developers can access over 1600 third-party models on Azure AI Studio.

Powerful Development Tools

Microsoft has released a series of new tools aimed at improving productivity for developers building AI applications. GitHub Copilot, as the most widely adopted AI development tool, has gained more capabilities through Copilot extensions. Microsoft has established an exciting new partnership with Cognition, introducing their "AI software engineer" Devin to the developer community. WebNN support brings high-performance AI capabilities to web applications, and Dev Drive accelerates workflows with near-instant file copying operations. Windows Subsystem for Linux (WSL) has added enterprise-grade security features, and PyTorch now has native support on Windows through DirectML.

Beyond Windows

Finally, the Windows ecosystem is surpassing the limitations of personal computers, enabling AI-driven experiences on various forms and devices. Windows 365 brings the powerful capabilities of cloud PCs to any device, and the new partnership with Meta allows Windows applications to extend into 3D space on Quest devices through the Volumetric API. This expanding ecosystem provides developers with more avenues to reach more users with AI applications built on the Microsoft platform.

In conclusion, the most valuable aspect of Microsoft's Copilot platform is its stability, thanks to its abstraction of underlying models and infrastructure layers. This abstraction allows developers to focus on building intelligent solutions without worrying about the complexity of rapidly evolving AI technology. This stability also translates into a consistent experience for end users, who can rely on Copilot-driven solutions to seamlessly run on different devices and platforms.