Technology

Add as a preferred source on Google

MediaTek Expects US$2.5 Billion From First AI Accelerator Program in Q4 2026

CEO Rick Tsai said the program could bring in about US$16 billion in 2027, or roughly 20% of the AI accelerator ASIC market.

Updated 2 min read
Investors face a presenter at a conference hall in Hsinchu, Taiwan. / EconoTimes.ai
Investors face a presenter at a conference hall in Hsinchu, Taiwan. / EconoTimes.ai

MediaTek expects its first data-center AI accelerator program to generate US$2.5 billion in fourth-quarter 2026 revenue, setting an early benchmark for the company’s expansion into AI chips.

At MediaTek’s Oct. 30 investor conference, CEO Rick Tsai said initial shipments began in September. He forecast about US$16 billion in program revenue in 2027, at the high end of the company’s target range. Tsai said that amount would represent roughly 20% of the total AI accelerator application-specific integrated circuit (ASIC) market.

In July, MediaTek set a target of 15% to 20% of the AI accelerator ASIC market in 2027, which it estimated at US$80 billion. The October revenue forecast puts the program at the upper end of that earlier share objective.

On the Oct. 30 call, MediaTek cited strong customer demand and its supply outlook, saying it was working with supply-chain partners to secure additional capacity.

MediaTek reported third-quarter consolidated revenue of NT$168.1 billion, up 10.4% from the previous quarter and 18.3% from a year earlier. Gross margin was 46.3%. For the fourth quarter, the company forecast consolidated revenue of NT$210.1 billion to NT$221.8 billion, an increase of 25% to 32% quarter over quarter, assuming an exchange rate of NT$31.7 per US$1. It projected gross margin of 45%, plus or minus 1.5 percentage points. The consolidated outlook covers companywide revenue, while the US$2.5 billion estimate applies to the first AI accelerator program.

MediaTek said its second AI accelerator project is on track for high-volume production in the first quarter of 2028. The company described it as designed for next-generation AI infrastructure, with the aim of delivering several times higher performance per watt and significantly lower total cost of ownership.

Related articles

View all