Recent interactions between many current and former executives on social platforms show that Microsoft is relaunching an internal project that was shelved many years ago. The goal is to reconstruct the performance and resource usage of Windows 11 from the bottom up, including significantly reducing the memory usage when the system is idle and the disk usage after a new installation.
Mikhail Parakhin, the former head of Windows and Bing, revealed on When he responded to current Windows Division President Pavan Davuluri's post about "improving the quality of Windows," he said that Davuluri was "restarting" the direction he and current Microsoft CTO Jeff Johnson promoted.
According to the target for that year, if the "20/20" project is successfully implemented, the memory footprint of Windows 11 in idle state is expected to be compressed to about 4.8GB, but Parashin also admitted that "we have never really finished it." Now fast forward to 2026, Microsoft once again emphasized in its latest communication that it will improve the performance, response speed and memory efficiency of Windows 11. The problems to be solved are still problems that have existed many years ago but have always been difficult to cure. This also raises questions from the outside world: Since Microsoft failed to complete such a seemingly basic optimization work back then, what changes have occurred now, and whether this round of optimization in 2026 can really be implemented, instead of just an attempt with big thunder and small rain.
Judging from Microsoft's own explanation, the reason why Windows 11 has high memory usage is first of all due to its huge background service system. Compared with previous versions, Windows 11 runs more components that reside in the background: telemetry and data collection, search indexing, security protection, widgets, and information flows will continue to be active in the background, and cloud synchronization services such as OneDrive are also constantly reading and writing data. The benefits of these designs are that a large number of functions are preloaded and pre-indexed, resulting in faster response when users click or call, and a smoother subjective experience. However, the price is that the basic memory usage of the system is increased, and the system continues to work even if it is seemingly "idle".
What really amplifies the problem are changes in the application ecosystem on top of the operating system. A large number of popular desktop applications use the Chromium-based Electron framework, or rely on Windows' built-in WebView2 technology to build interfaces and logic. WhatsApp desktop version and Discord are typical representatives. These applications often each carry a set of browser kernels, coupled with multi-process rendering, script execution and background tasks. Just one program may occupy hundreds of MB of memory, not to mention the common multi-application parallel usage scenarios for users. Even Microsoft's own applications such as Teams, Clipchamp, and Widgets components in Windows 11 also widely use WebView2, which further pushes up the overall memory usage level.
At the interface layer, Windows 11’s long-term reliance on a “spliced” technology stack also increases resource overhead. The current system uses a mix of traditional Win32 elements, UWP components, modern WinUI layers, and web interfaces built on technologies such as WebView2 and React. This hybrid model provides Microsoft with flexibility in the process of feature iteration and migration, but different parts rely on different rendering pipelines and system resources, which will inevitably bring additional performance and memory costs. Microsoft has publicly acknowledged this problem and confirmed that it is migrating more components to native frameworks such as WinUI 3 in order to reduce delays and improve efficiency. However, this process requires rewriting the core part of the system, which is costly in time.
As for the reason why the original "20/20" project aborted, Parashin did not give details, but it is not difficult to infer from outside observation that the key lies in the issue of resources and trade-offs. To substantially reduce the memory footprint of Windows, it means that background services must be deeply adjusted or streamlined, the UI technology stack must be streamlined, and even the expansion of web components in the system must be limited. These all involve fundamental changes at the architectural level. However, over the past few years, Microsoft has continued to add new features to Windows and strengthen integration with cloud services. On the other hand, it has vigorously embedded AI experiences into the system. These expansions themselves will bring more resource overhead. On the premise that platform capabilities continue to expand, it is almost impossible to "radically reduce system overhead" at the same time. Therefore, "20/20" is likely to be marginalized under realistic trade-offs and ultimately end in vain.
Unlike the internal attempts of the year, Microsoft this time chose to make a clear commitment through public channels: it will reduce the system's baseline memory footprint in future Windows 11 updates, free up more available RAM for applications, and provide a perceptible improvement in the smoothness of daily use. At the same time, the company will also focus on optimizing the responsiveness in high-load scenarios. The goal is to maintain a smooth and consistent experience of system interaction when multiple applications are running in parallel or even under heavy load, instead of being obviously "stuck" after opening several large applications. This includes improving multi-tasking switching, making switching between applications closer to "instantaneous" and reducing user waiting time.
On the technical path, Microsoft emphasized that it will continue to reduce interaction delays, optimize shared UI infrastructure, and accelerate the migration to native frameworks such as WinUI 3. By unifying and simplifying the interface layer technology stack and reducing the existence of duplicate and redundant components, the overall resource usage of the system is expected to be further tightened. Combined with the previous adjustments to the "de-Webization" of key interface components such as the Start menu, it can be seen that Microsoft is trying to replace the past implementation that relied on the Web with more native and lightweight technologies.
Changes in the external environment have also pushed Microsoft to seriously face Windows 11 performance issues to a certain extent. In recent years, criticisms surrounding Windows performance, resource usage, and user experience have escalated from small-circle technical discussions to mainstream public opinion topics, and it is no longer easy for Microsoft to turn a blind eye to this. The pressure on hardware and competitive products is equally significant: Apple’s emphasis on energy efficiency and resource utilization on its self-developed chips has redefined users’ expectations for desktop systems to be “power-saving, lightweight, and efficient.” Products such as MacBook Neo have also pushed memory usage issues into the spotlight. At the same time, global memory prices are rising, making memory resources more cost-sensitive, further strengthening the business urgency of optimizing Windows 11 performance.
In this context, user demands, market competition, and Microsoft's internal priorities can be said to rarely point in the same direction: to return Windows 11 to a more efficient and restrained resource utilization form. The unfinished tasks of "20/20" have now been put on the table again in another way. Whether this round of adjustments in 2026 can truly be implemented will ultimately be tested by time and user experience.