NVIDIA is expanding its AI empire in every direction. The company announced a broader partnership with MediaTek, full production for its Groq 3 LPX inference accelerator, and an Amazon deal adding 2 million NVIDIA GPU chips to AWS data centers.
The MediaTek agreement covers AI infrastructure, local AI computing, and automotive platforms. NVIDIA also invested $3.5 billion in convertible bonds issued by MediaTek, turning a technology partnership into a financial one—because apparently chips were not enough paperwork.
MediaTek will adopt NVIDIA’s NVLink Fusion platform, which NVIDIA describes as a prevalidated path for hyperscalers, cloud service providers, and frontier model developers to build custom XPUs. Within that ecosystem, MediaTek will help develop custom AI infrastructure while bringing its system-on-chip expertise to the broader platform.
One partnership, three computing fronts
The companies will continue working together on NVIDIA RTX Spark and DGX Spark PC chips, extending their collaboration into local AI computing. They will also develop platforms for AI-powered, software-defined vehicles, adding automotive systems to a partnership already aimed at data centers and PCs.
Jensen Huang, founder and CEO of NVIDIA, framed the strategy around AI’s spread across machines: “AI is transforming every computing platform — from the world’s largest AI factories to the PC and the car.” He also described MediaTek as “one of the world’s great semiconductor companies,” citing its expertise in system-on-chip design, connectivity, performance, and power efficiency.
That combination gives NVIDIA access to a company built around compact, integrated computing platforms, while MediaTek gains a role inside NVLink Fusion and NVIDIA’s expanding AI infrastructure stack. The plan is not one product; it is a shared route from large AI systems to devices and vehicles.
Inference capacity meets an AWS spending surge
NVIDIA Groq 3 LPX has entered full production, with Nebius becoming the first AI cloud to adopt the accelerator. In Artificial Analysis benchmarking running Gemma 4 31B, Groq 3 LPX delivered 3,400 output tokens per second and enabled four-times faster responsiveness for agents and latency-sensitive workloads than the nearest alternative platform.
Danila Shtan explained the target in plain terms: “Generation is the phase of inference that determines how responsive an AI system actually is, and that’s exactly what NVIDIA Groq 3 LPX is built to accelerate.” For AI agents, faster output is not decoration; it determines how long users wait for a system to complete a task.
Amazon is also expanding its NVIDIA relationship, agreeing to add another 2 million NVIDIA GPU chips to its data centers. The chips include NVIDIA Blackwell Ultra, Rubin, and Rubin Ultra GPUs, with deliveries headed to AWS data centers in 2027 and 2028.
Amazon has already agreed to deploy more than 1 million NVIDIA GPUs across AWS starting this year, but demand has exceeded expectations since the initial GPU deployment agreement. The expanded deal is worth tens of billions of dollars, a phrase that has become strangely routine in AI infrastructure announcements.
NVIDIA technology will span AWS beyond GPUs. Amazon plans to integrate NVIDIA networking hardware, open models, CPUs, data processing software, and robotics platforms across AWS, while NVIDIA will send an unspecified number of Vera CPUs to Amazon—some integrated with Rubin and others operating as standalone chips.
Amazon will also adopt NVIDIA’s full physical AI stack for its robots, including Omniverse, Cosmos, Isaac, and Jetson. At the same time, Amazon is investing in its own AI chips, including Trainium and Arm-built Graviton CPUs; its AI chip business has crossed a $25 billion annualized revenue run rate, driven by $225 billion in total commitments from AI labs.
The financial backdrop explains the scale. NVIDIA reported $96.2 billion in second-quarter sales, including $89 billion in data center revenue, up 117% from a year ago, and expects third-quarter revenue to reach $108 billion.
Huang called inference “the growth engine of AI” and said AI is generating profitable tokens: “If we had more compute, we could generate more profitable tokens, which results in more profit for all of the services.” NVIDIA, MediaTek, and Amazon are now building toward that equation across infrastructure, personal devices, vehicles, and robots.
Based on
- NVIDIA and MediaTek Expand Partnership to Build AI Edge-to-Cloud Platforms — unite.ai
- NVIDIA Groq 3 LPX Now in Full Production With World-Class Speed for Agentic AI | Markets Insider — markets.businessinsider.com
- Nvidia’s AI advantage is moving beyond the GPU | TechCrunch — techcrunch.com
- Amazon just tripled its order of Nvidia chips over ‘surging demand’ | TechCrunch — techcrunch.com
