The Rise of Cloud Computing: From Moore's Law to Alibaba's Shenlong Architecture

Recent discussions suggest that Moore's Law is reaching its limits, prompting the industry to look for new ways to sustain computing performance growth, with cloud computing emerging as a key solution.

In 2019, the global cloud market surpassed the $100 billion mark, driven by major providers such as Amazon, Microsoft, and Alibaba Cloud, indicating substantial future growth potential.

Cloud computing relies fundamentally on virtualization, a concept first proposed in a 1974 paper. While virtualization enables resource pooling and higher utilization, its overhead can be significant, leading to performance penalties—for example, a virtual CPU core may run at roughly half the speed of a physical core.

Early virtualization breakthroughs came from VMware in 1997, but traditional virtual machines often suffer from latency and reduced throughput, and shared resources raise concerns about security and fairness.

Alibaba Cloud addresses these issues with its self‑developed Shenlong architecture, a tightly integrated hardware‑software virtualization solution that minimizes overhead, delivers elastic VM experiences, and dramatically improves performance.

Shenlong achieves this by offloading compute‑intensive virtualization tasks to specialized hardware such as FPGA and ASIC accelerators, allowing CPUs to focus on pure computation. As a result, Alibaba Cloud reports up to 500× network and 2000× storage performance improvements compared to a decade ago.

The architecture not only boosts raw performance but also enhances resource isolation and data security, positioning it as a next‑generation platform for heterogeneous computing, NPU, and reconfigurable workloads.

Overall, the article argues that the future of computing will be defined by close hardware‑software co‑design, with cloud providers like Alibaba leading the way by turning hardware into a flexible, performance‑driving component rather than a static substrate.