Lisa su pitches AMD as China’s alternative to NVIDIA

AMD used its Shanghai developer event to position China as central to its roadmap and to court developers looking for an alternative to NVIDIA’s CUDA ecosystem. The strategy focuses less on headline chip specs and more on migration support, open-source tools, and long-term bets on the next wave of Artificial Intelligence applications.

Lisa Su brought AMD’s Advancing Artificial Intelligence developer event to China for the first time, signaling a sharper push into a market where NVIDIA’s position has weakened under export controls. In Shanghai, she said that in the next five years, five billion people will use Artificial Intelligence every day, and she framed China as being at the core of AMD’s roadmap. The move came shortly after Jensen Huang’s China trip, at a time when there was still no clear statement regarding the implementation of the H200 in China.

AMD’s challenge is not raw hardware alone but the software ecosystem built around CUDA. The company has competitive products, including the MI300X single – card with 192GB of HBM3 memory, but developers remain deeply tied to NVIDIA’s tools and workflows. AMD’s ROCm stack exists as an open-source alternative, yet moving from CUDA to ROCm requires code changes, engineering retraining, and support during the difficult first phase of adoption. Lisa Su’s strategy in China centered on that reality. Her keynote lasted only about ten minutes, while the rest of the event focused on technical workshops, migration guidance, framework optimization, and direct exchanges with engineers designed to show that work now done on CUDA can be shifted to ROCm.

The company sees China as a market where it can be chosen as a primary platform rather than as a backup supplier. When Lisa Su took over AMD in 2014, the company’s stock price was less than ?, and she rebuilt its competitiveness through Zen architecture and chiplet packaging. In six years, AMD’s share in the server CPU market increased from nearly zero to over 30%. In Artificial Intelligence chips, AMD has won notable customers, but the article describes those purchases as a hedge against NVIDIA rather than a sign of dependence on AMD. China offers a different opening because new buyers are evaluating alternatives, especially in emerging Agent applications where computing needs differ from large-model training and where developers are described as more open-source oriented and price-sensitive.

AMD’s position in China also reflects geopolitics. Since 2022, the US government has tightened export controls on Artificial Intelligence chips to China round by round. NVIDIA once held a 95% market share in China, and Jensen Huang said in May this year that this figure is now close to zero. AMD’s export product for China, the MI308, has obtained partial US licenses, and there are market rumors that a large domestic company is negotiating to purchase 50,000 MI308 chips. AMD has also put its 4,000 – strong R & D team in China on the front line and established four Artificial Intelligence Excellence Centers in China.

The weakness in AMD’s pitch is that its long-term vision is not yet matched by a China-ready flagship product. Lisa Su promoted a CPU + GPU dual – engine approach for the Agent era, arguing that combined architectures are better suited to handling logic, scheduling, and inference together. AMD’s advantage is that it has an x86 CPU, an Artificial Intelligence GPU, and the MI300A that integrates both on one chip, while NVIDIA’s CPU is based on Arm architecture and is described as being incompatible with more than 90% of the x86 software stacks in Chinese data centers. But the MI308 available in China is not an integrated product. It is a down – graded version of the MI300X, and the MI300A currently has no China – compliant version. AMD is therefore using 2026 to build relationships and ecosystem support, while betting that broader Agent demand between 2027 and 2028 will align with a compliant next-generation platform.

68

Impact Score

DeepWeb-Bench tests limits of deep research models

DeepWeb-Bench is positioned as a tougher benchmark for evaluating whether frontier language models can handle real deep research tasks beyond existing tests. Results point to derivation and calibration, rather than retrieval, as the main weaknesses in current Artificial Intelligence systems.

Google adds conversational ads to Artificial Intelligence mode

Google is rolling out new ad features in Artificial Intelligence Mode aimed at helping businesses, especially smaller advertisers, appear in generative search experiences. The additions bring conversational responses, recommended business listings and lead-generation tools directly into search interactions.

Google shifts its scientific Artificial Intelligence focus

Google is presenting a broader vision for scientific Artificial Intelligence that leans more heavily on agentic, general-purpose systems while still maintaining specialized tools. The shift suggests changing priorities in how the company sees Artificial Intelligence contributing to research.

General-purpose Artificial Intelligence tackles open math problem

OpenAI’s GPT-5 reportedly helped mathematician Ernest Ryu solve a long-standing convex optimization problem now under formal verification. The result points to a broader shift from Artificial Intelligence as a math assistant toward Artificial Intelligence as a source of original research.

Suno faces lawsuit from The American Dollar

Suno has been sued by production duo The American Dollar, who allege its generative Artificial Intelligence music service has damaged their sync licensing business. The case could sharpen the fair use fight around whether Artificial Intelligence training causes direct market harm.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.