News

[News] AI Company Astera Labs Expands in Taiwan, Local ODMs Expected to Benefit


2024-05-21 Semiconductors editor

Astera Labs, a leading provider of AI server connectivity solutions, has announced that it will gather Taiwanese manufacturers to establish its first Cloud-Scale Interop Lab outside of Silicon Valley in Taiwan. According to a report from Commercial Times, the company will closely collaborate with major Taiwanese ODM clients, while key manufacturers such as Quanta, Inventec, Wistron, Wiwynn, and Foxconn are expected to benefit from this initiative.

The emerging AI company Astera Labs has surpassed a market value of USD 10 billion and is renowned for providing high-speed transmission interface solutions for AI servers. Founded in 2017, the company celebrated its new public listing on NASDAQ this March.

Per a report from Business Today, the company, headquartered in California, USA, specializes in Retimer chips used for transmission in cloud data centers. These chips mitigate electronic signal attenuation issues, making them widely adopted in the market following PCIe Gen 5.

In response to the rapid expansion of the AI server market, Astera Labs is following in NVIDIA’s footsteps by establishing an R&D center, the Cloud-Scale Interop Lab, in Taiwan.

The report from Business Today further addresses that, according to Astera Labs’ financial reports last year, 60% of the company’s revenue came from Taiwan. Sanjay Gajendra, President and Chief Operating Officer of Astera Labs, stated that most of the company’s clients are major server ODMs based in Taiwan. In addition to server ODMs, TSMC is also an important partner for Astera Labs.

Sanjay pointed out that TSMC was an early investor in Astera Labs, and the company’s chips are all manufactured using TSMC’s cutting-edge processes. He also revealed plans to meet with TSMC’s CFO during this visit.

Sanjay Gajendra emphasized that the company will quickly expand its team, using Taiwan as a base in the Asia-Pacific region to support the PCIe 6.x test suite. This initiative aims to help businesses rapidly track and deploy solutions, enabling customers to integrate Aries 6 and achieve the industry’s lowest power consumption for PCIe 6.x and CXL 3.x Retimers.

NVIDIA’s next-generation GPU power consumption will reach 1400 watts. Sanjay Gajendra revealed that Astera’s technology is fully integrated into AI servers. As chip designs become increasingly complex, PCIe 6 achieves rapid data transmission for chips and can also connect GPUs across multiple racks.

In response to Astera Labs’ expansion in Taiwan, as per a report from TechNews, the aforementioned partners, including Quanta, Inventec, Wistron, Wiwynn, and Foxconn, have expressed their anticipation for this development. Foxconn has stated that it looks forward to continued collaboration with Astera Labs, fully utilizing the rigorously tested and field-validated PCIe/CXL Retimer solutions in its systems.

Quanta highlighted that the powerful Aries 6 Retimers, tested at the newly established Cloud-Scale Interop Lab in Taiwan, will enhance the promotion of reliable PCIe 6.x connectivity in next-generation AI and cloud infrastructure. Inventec, Wistron, and Wiwynn also remarked that the collaboration between both parties will continue to strengthen with the establishment of the R&D center in Taiwan.

Read more

(Photo credit: Astera Labs)

Please note that this article cites information from Commercial Times, Business Today and TechNews.

Get in touch with us