MediaTek, a prominent Taiwanese semiconductor company specializing in chip design, experienced a remarkable surge in its stock price, climbing more than 20% during the past week and achieving its strongest weekly performance since 2002. This impressive rally was triggered by emerging reports confirming a strategic partnership with Google on the development and design of tensor processing units (TPUs), which are specialized chips optimized for artificial intelligence workloads. The shares advanced steadily across five consecutive trading sessions on the Taipei exchange, concluding on a positive note last Friday, as market enthusiasm grew around MediaTek’s potential to carve out a significant niche in the rapidly expanding AI hardware landscape.
The catalyst for this excitement traces back to Google’s unveiling of its advanced Gemini 3 AI model on November 17, which showcased the model’s capabilities in handling complex AI tasks and underscored the viability of alternative chip architectures to those offered by industry leader Nvidia. Under the terms of the collaboration, MediaTek is taking on a key role in designing components of the next-generation TPUs, including input/output modules that facilitate communication between processors and other system elements, while production is expected to ramp up at Taiwan Semiconductor Manufacturing Company (TSMC) beginning next year. This partnership not only helps Google diversify its supply chain away from exclusive reliance on previous collaborators like Broadcom, but also leverages MediaTek’s established expertise in cost-efficient chip fabrication and its deep integration with TSMC’s advanced 3nm manufacturing processes, which promise enhanced power efficiency and performance for AI inference tasks.
Furthermore, the initiative reflects a broader industry shift toward custom application-specific integrated circuits (ASICs) tailored for AI, where TPUs excel in accelerating machine learning operations like those used in Google’s TensorFlow framework. By involving MediaTek, Google aims to reduce per-chip costs compared to prior designs, making these AI accelerators more accessible for both internal research and cloud customers facing surging demand for compute resources. This development positions MediaTek to extend Google’s TPUs beyond data centers into consumer devices, such as smartphones and edge computing platforms, where on-device AI processing is increasingly vital for privacy and real-time applications.
Analysts Boost Revenue Forecasts
Wall Street analysts expressed robust optimism about MediaTek’s trajectory this week, with several major firms elevating their ratings and financial projections in light of the Google partnership. For instance, analysts Charlie Chan and Daniel Yen at Morgan Stanley upgraded MediaTek’s stock from equal-weight to overweight status on Thursday, emphasizing that the upside from TPU contributions could effectively counterbalance the ongoing challenges in China’s smartphone sector, where demand has softened due to economic pressures and market saturation.
In a similarly bullish move, UBS analysts under the leadership of Sunny Lin significantly increased their revenue forecast for MediaTek’s TPU-related business, projecting $4 billion in sales by 2027—more than double their prior estimate of $1.8 billion. They further anticipate that these AI chips could represent around 20% of MediaTek’s total operating profit by 2028, provided the collaboration executes smoothly and capitalizes on the growing adoption of AI-optimized hardware in diverse applications. This projection aligns with MediaTek’s heavy investment in research and development, which stands at 23.3% of its budget, fueling innovations like the Dimensity 9400 chipset with 68 TOPS of neural processing power. Adding to the positive sentiment, reports surfaced that Meta Platforms is engaged in advanced discussions to integrate Google TPUs into its expansive data center infrastructure starting in 2027, with potential expenditures reaching billions of dollars on these chips.
Such a deal could unlock additional opportunities for MediaTek in the realm of custom ASICs, as it would validate the scalability of the TPU ecosystem and encourage other tech giants to pursue similar tailored solutions for their AI needs. Analysts note that this momentum is part of a larger trend, with AI ASIC adoption in data centers projected to reach 37% for inference workloads in 2025, shifting focus from general-purpose GPUs toward specialized designs.
AI Chips Offer New Growth Avenue
The partnership with Google signals a critical strategic evolution for MediaTek, which has been grappling with headwinds in its foundational business segments, including a notable decline in global smartphone demand and resultant margin pressures from competitive pricing. In particular, China’s smartphone market, a key revenue driver for MediaTek, contracted by 3% year-over-year during the third quarter of 2025, according to detailed insights from market research firm Omdia, reflecting broader consumer caution amid economic slowdowns and the maturation of the mobile device sector.
MediaTek’s CEO, Rick Tsai, highlighted the company’s forward-looking ambitions during an October earnings call, stating that it anticipates generating $1 billion in revenue from cloud AI chips as early as 2026, with the inaugural Google project poised to deliver multiple billions in additional income by 2027. Tsai outlined an aggressive goal for MediaTek to secure 10% to 15% market share in the burgeoning $50 billion data center ASIC chip market within the next two years, capitalizing on its ability to deliver high-performance, energy-efficient solutions at competitive prices. This vision is bolstered by MediaTek’s ongoing collaborations, such as co-designing the energy-efficient GB10 Grace Blackwell Superchip with Nvidia and embedding TPUs into consumer ecosystems, which collectively diversify its portfolio beyond traditional mobile processors.
On a broader scale, this AI pivot comes at a pivotal time for the semiconductor industry, where companies like Google are investing heavily—up to $75 billion in capital expenditures for 2025, with half allocated to cloud infrastructure—to maintain a competitive edge against rivals including Microsoft and OpenAI. By partnering with MediaTek, Google not only mitigates risks associated with supply chain dependencies but also advances innovations like the Gemma 3 AI model, which runs efficiently on single GPUs or TPUs, enabling lighter and faster AI deployments across various scenarios. For MediaTek, this translates to embedding itself deeper into the AI supply chain, from data center servers to edge devices, potentially transforming it from a smartphone-centric player into a multifaceted AI enabler.
Wall Street’s overall consensus remains firmly positive, featuring 23 buy recommendations, 10 holds, and zero sell ratings from covering analysts, reflecting confidence in MediaTek’s execution capabilities. The average price target implies an additional 9% potential increase over the next 12 months, providing a clear path for further appreciation. Despite the recent weekly surge, MediaTek’s shares are still trailing by about 2% to 3% for the year-to-date period, indicating substantial recovery potential as AI demand continues to accelerate globally and positions the company for sustained growth in this high-stakes technology frontier.
