What’s stopping Asia’s chipmakers from taking on Nvidia? The answer lies in this bottleneck
Foundry access is just one constraint that startups are facing
Nvidia's dominance in the semiconductor industry has created a wave of AI chip startups across Asia, many founded by former employees of Intel and AMD. However, these startups face a significant bottleneck in accessing advanced fabrication facilities, such as those operated by TSMC and Samsung Foundry. These facilities are few and far between, making it challenging for startups to secure a manufacturing partner.
TSMC, the market leader in advanced nodes, is still a major obstacle for startups looking to enter the AI chip market. Getting a production allocation at TSMC is not an easy feat, as the company evaluates each request carefully. To make matters worse, advanced nodes like 3 nm processes are incredibly difficult and expensive to manufacture, adding to the challenge for startups.
Despite these hurdles, investors continue to pour funds into AI chip startups. For example, Bengaluru-based Agrani Labs, with a team of former Intel and AMD executives, has raised $8 million so far and is reportedly in talks for more than $100 million. However, even with significant funding and strong talent, securing TSMC's production capacity remains a significant challenge.
One potential solution to this bottleneck is advanced packaging solutions like TSMC's CoWoS and Silicon Box's panel-level packaging. These advanced packaging techniques combine AI processors, high-bandwidth memory (HBM), and other components into a single chip package, making it easier for startups to compete in the market. However, TSMC's limited capacity for CoWoS production also poses a challenge for startups trying to secure this advanced packaging.
Written by urgent.news from The Business Times - Companies & Markets's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
