MediaTek says it is on track to generate $1B in revenue from its data center AI accelerator ASIC chips in 2026, after reporting Q3 revenue up 7.8% YoY to $4.64B
Taiwan's top chip design company, MediaTek (2454.TW), expects to earn revenue of billions of dollars from its AI accelerator ASIC chips by 2027 …
Context & Ripple Effects
MediaTek had already returned to growth, with 2024 sales and profit growth following its earlier first year-over-year quarterly improvement. This update identifies data-center AI accelerators as a prospective new revenue stream alongside that recovery.
The target places MediaTek in a custom-AI-chip market where Broadcom had already projected $10B in AI-chip revenue for 2024. It also aligns with wider Taiwanese chip demand that had lifted TSMC's 2024 revenue.
First-order effects
- MediaTek is setting a measurable 2026 revenue objective for its data-center AI accelerator ASIC business, making the segment a material strategic focus rather than an adjacent experiment.
- The company’s reported quarterly growth is now paired with an expectation that ASICs can contribute billions of dollars by 2027, raising the importance of execution in this product line.
Second-order effects
- The stated target strengthens pressure on ASIC vendors, including larger incumbents, to compete for data-center AI designs on performance, customer support and delivery rather than relying solely on general-purpose accelerators.
- If MediaTek converts the target into shipments, demand would extend further through Taiwan’s chip-design and manufacturing chain, consistent with AI demand already supporting TSMC's revenue growth.
Third-order effects
- The development supports a shift toward a more segmented AI-compute market, in which custom accelerators can become a meaningful complement to merchant AI chips for particular workloads and customers.
- Whether that shift broadens depends on sustained customer adoption and MediaTek’s ability to execute at data-center scale; a revenue target alone does not establish either.
The trend: AI infrastructure spending is widening the pool of chip designers pursuing custom data-center accelerators, transmitting demand beyond the leading general-purpose AI-chip suppliers.