Huawei, OpenAI & Tencent Reveal Three Fronts In The Global Chip Race

The global chip race is getting more complicated and, ironically, more creative.

Huawei is redesigning chip architecture because advanced capacity remains tight, Tencent is reportedly leasing 100,000 AI chips overseas, and OpenAI is working with Synopsys to use AI to speed up chip design.

Together, the developments show that semiconductor competition is no longer only about manufacturing better processors. Access to compute, new chip architectures, and faster design cycles are becoming equally important.
 

TL;DR

 

• Huawei’s Mate 90 series uses its Kirin 9050 Pro and LogicFolding architecture as the company continues working around limited access to advanced semiconductor capacity.
• Tencent reportedly secured access to around 100,000 advanced AI chips through a five-year Oracle lease estimated at $7 billion.
• OpenAI and Synopsys are developing GPT-Synopsys to help engineers optimize chip designs and potentially cut weeks or months from development cycles.

 

Huawei Uses LogicFolding To Push Mate 90 Performance Despite Chip Constraints

Huawei has unveiled its Mate 90 smartphone series with another attempt to squeeze more performance from technology available under continued semiconductor restrictions.

Premium models, including the Mate 90 Pro Max, use Huawei's Kirin 9050 Pro system-on-chip and a technique called LogicFolding. Instead of relying only on improvements from smaller manufacturing processes, LogicFolding restructures internal chip wiring in three dimensions to pack computing resources more densely, although the approach requires more wafers to manufacture.

Huawei consumer business chief Richard Yu acknowledged the underlying challenge, saying, “Advanced semiconductor capacity remains very limited in China, and Huawei's Ascend AI chips also draw on that same capacity.”

Chinese manufacturers still rely on deep ultraviolet lithography while working toward extreme ultraviolet technology.

Huawei says the top Mate 90 models deliver 31% higher overall performance than the previous generation. The phones also run HarmonyOS 7, its Android-free operating system, which Huawei says supports more than 450,000 apps and services.

That is one response to chip scarcity. Tencent's reported strategy takes a very different route: securing computing power outside China.
 

 

Tencent Reportedly Leases 100,000 AI Chips From Oracle For Its AI Expansion

Tencent has reportedly signed its largest overseas chip-leasing agreement, giving the Chinese technology company access to around 100,000 advanced AI chips that are unavailable domestically.

The reported five-year arrangement spans multiple Oracle data centers in Southeast Asia and is estimated at roughly $7 billion, with an upfront payment of around 30%. The report has not been independently verified, while Oracle and Tencent did not provide comments on the matter.

The scale reflects Tencent's growing computing requirements as it invests in large language models and other AI products. It also highlights another pressure point in the chip race: companies do not necessarily have to own advanced processors if cloud infrastructure can provide access to them.

Meanwhile, we have been tracking how companies are exploring alternative paths through China's expanding AI chip ecosystem as access to advanced hardware becomes increasingly strategic.

Yet access is only one bottleneck. Another sits much earlier in the semiconductor lifecycle, where chips must first be designed.
 

OpenAI And Synopsys Bring Generative AI Into Semiconductor Chip Design

Synopsys and OpenAI have struck a deal to develop GPT-Synopsys, an AI model specifically trained for semiconductor design work.

The model will learn to use Synopsys software across processes ranging from describing chip circuits to determining how billions of transistors should be arranged on silicon. OpenAI co-founder Greg Brockman said the goal is “to be able to shave off weeks, months from the design process and to bring more chips to the world.”

OpenAI will receive a training subscription fee, while the companies will share revenue when customers use the system. Importantly, AI-generated work will still go through Synopsys' traditional verification tools to confirm that proposed designs satisfy physical requirements.

The partnership also arrives as OpenAI expands its broader involvement in semiconductor technology, including its own custom AI chip efforts.
 

 

Global Chip Race Expands From Manufacturing To Access, Architecture And AI Design

Taken together, Huawei, Tencent and OpenAI are approaching different constraints in the same semiconductor race.

Huawei is changing chip architecture to extract more performance from limited manufacturing capacity, Tencent is reportedly seeking advanced compute through overseas infrastructure, and OpenAI is trying to accelerate the design process itself.

The chip race, then, is increasingly being fought across the entire semiconductor stack, from how processors are designed and built to where companies can access enough computing power to keep their AI ambitions moving.