Alibaba’s AI Chip Gambit

Alibaba stock (NYSE:BABA) surged by close to 13% in Friday’s trading to about $135 per share. BABA stock also remains up by almost 60% year-to-date. The recent rally follows a positive Q1 earnings report which saw Alibaba’s cloud business sales surge. Moreover, there have been reports that Alibaba has developed a new AI chip for its cloud computing unit. Now, the new chip isn’t about to challenge Nvidia (NASDAQ:NVDA) Hopper series of chips, let alone the new Blackwell lineup, but that’s not really the point. It’s about making sure Alibaba can secure a supply of AI semiconductors despite U.S. export curbs, while also giving its cloud business a stronger edge as AI adoption keeps climbing.

Image by Bethany Drouin from Pixabay

What We Know About The Chip

Alibaba’s T-Heat unit has been working on AI chips for several years now. In 2019, the company introduced the Hanguang 800 aimed at conventional machine learning models.  However, with the new chip, Alibaba is stepping deeper into the AI hardware race working with large language and diffusion models that drive today’s generative AI technology. The new chip is designed specifically for inference workloads – the stage where trained models generate answers and recommendations in real time.

It’s expected to be built on a 7 nanometer process, making it much more capable and versatile than the older Hanguang chip. Early reports suggest it will be focused on workloads such as recommendation systems and natural language processing. Importantly, the chip is said to be compatible with Nvidia’s software ecosystem which could enable programmers to tweak and reuse existing code. How Snowflake Stock Surges To $450

This move also comes against a geopolitical backdrop. The U.S. has blocked leading-edge chip exports to China, with Nvidia’s H20 restricted earlier this year. Although shipments have since been conditionally allowed, Chinese firms still face uncertainty, and Beijing has reportedly told companies to avoid relying on the H20. By developing its own chip, Alibaba can reduce dependence on U.S. suppliers while catering to China’s fast-growing demand for AI capacity.

Why Inference Matters Now

AI’s early growth wave was driven by training massive models, a process that made Nvidia the clear winner as its GPUs became the gold standard. The company’s sales have surged, rising from $27 billion in FY’23 to an expected $206 billion in FY’26. But training is likely to be a front-loaded, compute-heavy process that could begin to show diminishing returns, with larger models not always delivering proportional accuracy gains. The availability of high-quality training data is also becoming a bottleneck, as much of what’s easily accessible on the Internet has already been absorbed into today’s models. Inference, by contrast, happens continuously in production. It’s lighter per task but scales across millions of users and applications. This is the segment Alibaba is targeting with its new chip. While inference chips don’t need the same raw power as Nvidia’s flagship training GPUs, they are still likely a key pillar of AI systems.

Alibaba’s Strategy

Unlike Nvidia, Alibaba will unlikely sell its chip directly to customers. Instead, it will use the hardware to power Alibaba Cloud, with customers renting computing power. This should deepen customer reliance on Alibaba’s ecosystem while helping to drive recurring revenues. Owning both the hardware and the cloud platform also means tighter integration, and this could also result in efficiency gains. Alibaba is backing this strategy with capital, pledging 380 billion yuan (about $53 billion) toward AI infrastructure over the next three years. The company has good reason to do this. Its cloud division grew 26% year-on-year last quarter, and AI-related revenue has posted triple-digit growth for eight consecutive quarters.

Overall, the new chips are likely to complement Nvidia’s GPUs in Alibaba’s broader AI strategy. The company will probably continue to rely on Nvidia hardware for training AI models in the near term, while its own chips focus on powering cloud-based inference at scale. Other Chinese players, too, are doubling down on AI chips. Baidu, Huawei, and startups like Cambricon are all building AI semiconductors. But Alibaba’s established scale in cloud computing gives it a distribution advantage. It can immediately roll new chips into its vast data centers and monetize them through its existing customer base.

The Trefis High Quality (HQ) Portfolio, with a collection of 30 stocks, has a track record of comfortably outperforming its benchmark that includes all 3 – S&P 500, Russell, and S&P midcap. Why is that? As a group, HQ Portfolio stocks provided better returns with less risk versus the benchmark index; less of a roller-coaster ride, as evident in HQ Portfolio performance metrics.

Invest with Trefis Market-Beating Portfolios

See all Trefis Price Estimates