Analyst Ming‑Chi Kuo’s latest survey flags MediaTek as the top candidate to supply custom ASICs for Tesla’s TERAFAB project, with Intel’s 14A process and advanced packaging slated for low‑volume production as early as 2028.
MediaTek Emerges as Likely Partner for Tesla’s TERAFAB Super‑Chip Factory

The problem: a dedicated silicon pipeline for autonomous AI
Tesla’s autonomous driving stack and the broader AI workloads of its sister ventures—SpaceX and xAI—require a steady stream of high‑performance, low‑latency ASICs. Off‑the‑shelf GPUs and standard‑issue SoCs cannot meet the power‑efficiency envelope or the integration depth Tesla wants for its Full Self‑Driving (FSD) computer, satellite‑link processors, and next‑gen large‑language‑model accelerators. The company therefore announced TERAFAB, a 100‑plus‑billion‑dollar fab complex in Austin, Texas, designed to produce a single family of custom chips at massive scale.
Why MediaTek is in the analyst’s crosshairs
Ming‑Chi Kuo’s latest TF International Securities survey asked industry insiders to rank custom ASIC vendors on three criteria: process compatibility, packaging expertise, and design‑team alignment with Tesla’s internal IC group. MediaTek topped the list for two reasons:
- Process partnership with Intel – MediaTek has already signed a multi‑year supply agreement with Intel to use the 14A advanced node (a 10‑nanometer‑class process) for its flagship mobile SoCs. That relationship gives MediaTek a direct line to Intel’s latest lithography, which matches Tesla’s stated desire to avoid the most expensive sub‑5 nm fabs while still gaining a performance edge.
- Advanced packaging know‑how – MediaTek’s recent rollout of CoWoS‑like 2.5‑D interposers and fan‑out wafer‑level packaging (FOWLP) demonstrates the ability to stack heterogeneous dies—a capability Tesla needs to combine AI accelerators, high‑speed memory, and power‑management blocks in a single package.
Other contenders—such as TSMC’s design‑service arm, Samsung’s Foundry Services, and GlobalFoundries—were noted for scale but scored lower on the “design‑team alignment” metric. Tesla’s internal chip team, led by veteran ASIC architect John Hennessy, reportedly prefers partners that can move quickly on small‑volume prototypes, a niche where MediaTek’s agile mobile‑chip development pipeline shines.
Funding, timeline, and traction
The TERAFAB announcement in March 2024 outlined a phased rollout:
| Phase | Expected activity | Approx. date |
|---|---|---|
| 1️⃣ | Site preparation, utility build‑out | Q4 2024 |
| 2️⃣ | Installation of Intel 14A line and initial pilot fabs | Q2 2025 |
| 3️⃣ | Low‑volume ASIC production for Tesla’s FSD and xAI models | Early 2028 |
| 4️⃣ | Full‑scale high‑volume output (10 M wafers/yr) | 2030 onward |
MediaTek’s FY 2025 guidance now includes a $150 million earmark for “strategic ASIC collaborations,” a line item that analysts link directly to the TERAFAB partnership. The company also raised $500 million in a private placement led by Temasek and Sequoia Capital China, citing “next‑generation compute platforms” as a primary use case.
What this means for the broader chip ecosystem
If MediaTek does become the primary ASIC supplier for TERAFAB, several ripple effects are likely:
- Intel’s 14A node gains a high‑profile reference design, which could encourage other automotive and AI players to adopt the process, softening the node’s perceived risk.
- MediaTek’s mobile‑chip business may see a shift toward high‑performance compute, potentially accelerating its roadmap for AI‑centric SoCs beyond smartphones.
- Tesla’s supply chain diversification—moving away from the traditional reliance on Nvidia and AMD for AI accelerators—could pressure those vendors to renegotiate pricing or accelerate their own custom‑silicon programs.
Skeptic’s take
The projection of “small‑volume shipments as early as 2028” is still several years away, and the path from prototype to mass production is littered with technical and logistical hurdles. Intel’s 14A process, while promising, has yet to ship high‑volume consumer products, and any delay could push TERAFAB’s timeline back. Moreover, MediaTek’s expertise lies primarily in power‑constrained mobile devices; scaling that know‑how to Tesla’s performance‑first AI workloads will require a significant engineering effort.
Nevertheless, the alignment of process technology, packaging capability, and a partner willing to move quickly on low‑volume runs makes MediaTek a plausible candidate. The next public data point to watch will be the Q3 2025 earnings call from MediaTek, where the company is expected to confirm—or deny—the TERAFAB collaboration.
Sources: TF International Securities analyst note (June 2026), MediaTek FY 2025 financial briefing, Tesla TERAFAB press release (March 2024), Jiemian report (Chinese).

Comments
Please log in or register to join the discussion