If you own a Samsung Galaxy Tab S12 or an Amazon Fire TV Stick 4K Max, you own a MediaTek chip.
Most people never notice the name, yet MediaTek ranks among the world’s largest chip designers. The company says its chips power more than 2 billion devices a year. According to Counterpoint Research, MediaTek is the largest supplier of smartphone processors by shipment volume. The company’s 2025 revenue reached a record $19.1 billion.
The industry has noticed, even if shoppers haven’t. In August, Nvidia invested $3.5 billion in MediaTek and expanded a partnership spanning PCs, data centers, and cars. At an analyst briefing I attended on October 21, MediaTek laid out how far it wants to take that momentum.
MediaTek is expanding into new AI categories, mostly through partners such as Nvidia and Google. The bigger prize is what it does next. Two emerging categories have plenty of room for new ideas: hybrid agentic AI computers, meaning laptops and phones that run AI agents on the device as well as in the cloud, and AI wearables. MediaTek’s technology portfolio and engineering put it in a strong position to lead in both. Here is why, and what it would take.
Why Phone Engineering Fits the AI Era
A smartphone chip combines general computing, graphics, AI processing, and camera processing on one piece of silicon. MediaTek also builds the modems, Wi-Fi, Bluetooth, and power-management chips around it.
The hard part is making it all work together inside a small power and heat budget. Take a video call with live translation. The phone must process camera and microphone input, hold a wireless connection, run the app, and do the AI work at once. How well the chip juggles those jobs determines whether the call stays smooth, how warm the phone gets, and how long the battery lasts.
AI brings the same pressure to laptops and data centers. Laptop makers want more AI in thin designs with all-day battery life. Data center operators pay for every watt. MediaTek has spent two decades solving this problem in phones, and it now has more places to sell the answer.
How Nvidia Opens New Doors
RTX Spark is the clearest example. Nvidia unveiled the Arm-based processor in June, and laptops from Asus, Dell, HP, Lenovo, Microsoft, and MSI are now on sale. The chip pairs an Nvidia Grace CPU, co-developed with MediaTek, with Nvidia Blackwell graphics. MediaTek helped develop the custom CPU design and contributed expertise in memory, connectivity, and power-efficient integration.
Each company brings different strengths. Nvidia supplies the graphics, its CUDA software platform, and its relationships with Microsoft and app developers. MediaTek supplies the engineering to fit processing, memory, and connectivity inside a thin laptop’s power and space limits. MediaTek gains a path into premium Windows PCs without rebuilding Nvidia’s software ecosystem.
The two companies have worked together on car platforms since 2023. Their computing work began with DGX Spark, a compact desktop AI system for developers built on the GB10 chip they co-designed. According to Nvidia, one unit with 128 GB of memory can run AI models with up to 200 billion parameters. For enterprise teams, it means building and testing AI agents at their desks, which can cut cloud spending during development.
In August, Nvidia backed the relationship with money, buying MediaTek convertible bonds, and the companies expanded their work across data centers, PCs, and cars. “MediaTek is one of the world’s great semiconductor companies, with exceptional expertise in system-on-chip design, connectivity, leading performance and power efficiency,” Nvidia CEO Jensen Huang said in announcing the deal.
Quiet Push Into AI Data Centers
Big cloud providers increasingly design their own AI chips to run their workloads at lower cost. These custom processors are often called XPUs. MediaTek wants to be the company that designs and delivers them.
At the briefing, MediaTek described two routes. In the first, it develops a customer’s chip using Nvidia’s NVLink Fusion platform, incorporating prequalified Nvidia connection and memory blocks. The chip is designed to integrate with Nvidia rack systems, reducing custom integration work and schedule risk.
In the second, MediaTek designs the whole chip around the customer’s workload. That route takes more engineering, so it pays off only when a workload is large and steady. Either way, the customer wants the same result: more AI requests handled within the same power budget, so each request costs less to serve.
MediaTek told investors in July that production of its first AI accelerator, built for an unnamed U.S. hyperscale customer, would begin this quarter. It expects data center revenue to exceed $2 billion in 2026. Its second accelerator is on track for high-volume production in 2028.
Broadcom and Marvell lead this market and hold deep customer relationships. MediaTek will earn repeat business only by shipping on time, securing supply, and proving operating savings.
Momentum in new categories makes MediaTek a stronger supplier. A breakout product would make MediaTek the first partner device makers seek out. To get there, MediaTek needs to lead with its partners rather than wait for them.
Three opportunities stand out: lower-priced AI laptops, phones that double as PCs, and AI wearables. Each one also must answer a hard question: why MediaTek and not Qualcomm, its main rival in phone chips?
Bringing RTX Spark to Laptops Under $1,500
RTX Spark laptops start at $2,599. That price puts them in premium territory for professional creators, developers, and gamers. Most Windows laptop buyers spend far less, and much of the market sits well below $1,500.
MediaTek should work with Nvidia on new RTX Spark laptop configurations under $2,000, ideally under $1,500.
The design should start with practical AI uses, such as document search, research help, and agents that run on the laptop instead of in the cloud. Then it should fit the graphics, memory, and CPU to a set price. Cut too deep, and the laptop loses the local AI the chip promises, so the real work is choosing what to keep. Memory is a particular cost pressure, since memory prices have climbed sharply this year.
Here, MediaTek’s edge over Qualcomm is Nvidia. Qualcomm’s Snapdragon X chips already power Windows AI laptops well under $1,000, but they don’t offer Nvidia graphics. For everyday buyers, the potential advantage is better performance in games and creative apps.
RTX Spark launched with support for hundreds of games, along with Adobe Photoshop and Premiere. For developers, another draw is CUDA, Nvidia’s platform for building and running AI models. A lower-priced RTX Spark would bring both to a price band where buyers can’t get them on Arm today.
App compatibility is the hurdle both companies share. Arm laptops run many older Windows apps and games through emulation unless developers ship native versions. MediaTek and Nvidia would need to show that buyers’ everyday apps run reliably and perform well.
Turning Android Phones Into PCs
Google is bringing Android apps to bigger screens through Googlebook and improving how phones work with external monitors. MediaTek told 9to5Google its new Dimensity CX C10 Max chip will debut in an upcoming Lenovo Googlebook. The first Googlebooks use Intel and Qualcomm chips, so MediaTek starts behind.
Adaptive apps make the bigger screen useful. An email app can show a single message on a phone, then the inbox and calendar side by side on a monitor. Add a keyboard, mouse, and resizable windows, and the phone becomes a workspace.
This direction connects with the Lotus concept, our independent SmartTech Research analysis of phone-powered computing. Lotus imagines a smartphone that docks into a desk station or a laptop-style shell. The phone stays the computer; the accessories supply the larger screen, full-size keyboard, and connections.
MediaTek could set itself apart from Qualcomm here by making the whole system easier to build. That means reference designs, tested docking connections, power management tuned for docked use, and shared accessory standards developed with phone makers and Google.
A manufacturer should be able to build a phone that conserves battery power on the go, then handles document editing and multitasking on a monitor without overheating. Scale gives MediaTek reach: as the largest supplier of smartphone processors, it can put one design in front of many phone brands at once. Googlebooks and docked phones would serve different buyers, but both gain from the same adaptive apps.
The hurdles are commercial as much as technical. Manufacturers that sell both phones and laptops may hesitate to promote a product that replaces some laptop purchases, and app developers still need to add the desktop features people expect.
MediaTek could borrow a page from Nvidia to close that gap. Nvidia built much of its AI lead by giving developers free tools, code libraries, and training that make its hardware easy to build on. MediaTek already offers on-device AI tools through its NeuroPilot platform.
It could go further and work with Google and independent Android developers on adaptive apps, starting with the makers of popular note-taking, photo-editing, and project-management apps, where support from a chip partner can still make a difference. A handful of great adaptive apps could do more for the category than a faster chip.
AI Glasses With the Phone as the Brain
At Mobile World Congress in March, MediaTek showed AI glasses connected to a phone running its Dimensity 9500 chip. The phone ran the AI model, which handled text, images, speech, and video, with some features working offline. MediaTek says moving the computing to the phone allows ultralight glasses.
Keeping the heavy processing in the phone reduces the heat and battery drain in the glasses. The trade-off is dependence on the phone and the wireless link. MediaTek’s modem, Wi-Fi, and Bluetooth expertise fits this problem well. If it can make the link fast and dependable enough for help to feel instant, it gives glasses makers a lighter design to build on.
Qualcomm is courting glasses makers too, through its Snapdragon START program, so MediaTek needs to move from demo to partner products. It hasn’t named a partner bringing the design to market. Comfort and clear privacy controls will help determine whether people wear them.
What Could Stall the Breakout?
The core business is under pressure. MediaTek’s mobile phone revenue fell 20% year over year in the second quarter, which the company blamed on weak demand as rising component costs, led by memory, pushed up smartphone prices.
The slump isn’t unique to MediaTek. Counterpoint Research says global smartphone chip shipments fell 15% in the first half of 2026, with MediaTek and Qualcomm each down more than 25%. Growth in connectivity, automotive, and TV chips carried MediaTek through the quarter.
In data centers, the test is delivery. A late chip disrupts a customer’s plans, and advanced packaging, memory supply, and factory capacity can limit output no matter how good the design. Custom work also ties big engineering bets to a few customers.
Nvidia lowers some of the burden but adds dependence on its technology and priorities. In the end, design wins count only if they become supported products that make money.
What to Watch
I’ll judge MediaTek’s progress against its own targets and the opportunities outlined here.
First, the production targets MediaTek has set:
- The first accelerator enters production this quarter, and the second remains on track for high-volume production in 2028.
Second, the moves I’d like to see:
- Lower-priced RTX Spark configurations reach laptops under $2,000, and ideally under $1,500, in 2027.
- Device makers beyond Lenovo adopt the Dimensity CX C10 Max for Googlebooks.
- MediaTek works with Google, device makers, and developers to turn its glasses demo and phone-to-PC ideas into products partners ship.
MediaTek already makes many familiar products better, but it needs to explain those benefits in plain terms. Qualcomm’s Snapdragon marketing offers a model: it connects chip capabilities to benefits buyers recognize, such as sharper photos, smoother graphics, and longer battery life.
Giving device makers a reason to promote the MediaTek name starts with showing buyers what its chips make possible in everyday use. Its breakout will come from delivering new kinds of products and better experiences at prices more people can afford.
