Wiwynn has initiated shipments of AI training servers, which currently only account for a small portion of the company's revenue but are projected to increase significantly by 2024, according to the Taiwan-based ODM.
During Wiwynn's recent earnings call, CFO Harry Chen pointed out that pandemic data center demand has come to an end.
Wiwynn's first-quarter consolidated revenue increased 46.4% on year, reaching NT$74.23 billion (US$2.42 billion). Net operating profit totaled NT$4.48 billion, net profit before tax NT$4.16 billion, and net profit after tax NT$3.3 billion. The increases are attributed to uneven supply within the supply chain returning to normal. Revenue and profit for the period set new company records. Inventory days also reached a two-year low.
Chen noted that Wiwynn gained a new customer in the first quarter, but did not disclose the name of the customer. He added that revenue from its top two customers, Microsoft and Meta, continues to grow. Industry sources named Amazon Web Services (AWS) as the new customer.
The other source of Wiwynn's first-quarter revenue momentum was AI. Chen stated that AI server shipments for some customers began in the first quarter.
Responding to concerns that AI servers will affect gross margins for downstream assemblers, Chen said Wiwynn saw no impact in the first quarter.
Even with increased AI server shipments in the first quarter, Chen said Wiwynn's gross margin remained the same as before and profits were not diluted.
Chen pointed out that the timeline for servers from development to shipment is around one year. As such, Wiwynn expects AI servers will account for a higher proportion of shipments in 2024.
If the proportion of AI server shipments increases in the future, it may affect profits, Chen said. If that happens, Wiwynn will work together with the supply chain and its customers to find the best mutually acceptable business model.
Industry sources explained that AI deep learning is split into two phases: training and inference. Customers typically use general cloud servers with accelerator cards for inference, while AI servers are mainly used for training.
According to Chen, the industry is optimistic about the long-term demand for cloud data centers. Wiwynn will focus capital expenditures on real estate infrastructure in 2023 as part of its long-term preparations. Investments in machinery and equipment will begin in the second half of 2023 or 2024.
Wiwynn recently approved a capital increase plan for its factory in Mexico, for a total transaction value of NT$1.23 billion. The increase will go toward expanding production capacity.
In the future, Wiwynn's Taiwan factory will focus on motherboard production with help from its Malaysia factory. The Mexico factory will focus on backend assembly.
Article translated by Eifeh Strom
