Nvidia CEO Jensen Huang will be the keynote speaker at CES 2025, sparking speculation that the company may unveil its next-generation RTX 50-series GPUs. While AI will likely be a central theme, the presence of RTX 50-series cards is highly probable, considering Nvidia’s history of using CES for graphics card launches.
ByteDance, the parent company of TikTok, is reportedly developing two new AI GPUs, one for training and one for inference, to reduce its reliance on NVIDIA. The new GPUs will be manufactured using TSMC’s 5nm process node and are expected to enter mass production in 2026. This move comes as ByteDance faces challenges with NVIDIA’s high prices and limited availability of AI GPUs.
Oracle’s Larry Ellison and Tesla’s Elon Musk recently appealed to Nvidia CEO Jensen Huang for more GPUs during a dinner in Palo Alto. The demand stems from the growing need for GPUs in AI applications, highlighting the importance of securing these chips in a market facing shortages.
Oracle’s ambitious plans for an AI supercluster powered by over 130,000 NVIDIA GPUs have led to a scramble for the company’s chips. Even Elon Musk, known for his Tesla and SpaceX ventures, was reportedly at the dinner table “begging” NVIDIA CEO Jensen Huang for more GPUs. The demand for NVIDIA’s AI hardware highlights the company’s dominant position in the market and the growing need for powerful chips to fuel the artificial intelligence revolution.
Rumors are swirling about Nvidia’s upcoming RTX 50-series GPUs, suggesting they’ll demand even more power than their predecessors. The flagship RTX 5090 could draw up to 600 watts, significantly higher than the RTX 4090’s 450 watts. While Nvidia hasn’t officially announced these cards, various sources point to increased power consumption across the lineup, impacting system requirements and potential energy usage.
Backprop, an Estonian GPU cloud startup, has discovered that a single NVIDIA RTX 3090 GPU can power an AI chatbot capable of handling customer service requests for hundreds of users simultaneously. This finding suggests that businesses can implement AI chatbots without needing vast GPU clusters.
Samsung Electro-Mechanics (SEMCO) has announced a partnership with AMD to provide high-performance substrates for the tech giant’s next-generation CPUs and GPUs. This collaboration focuses on the growing demand for high-performance computing in data centers, with Samsung investing heavily in advanced substrate technology to meet future needs.
Intel is set to take the stage at Computex 2024, and we’re expecting some big announcements. Here’s what to expect from the keynote, including new chip announcements, AI updates, and more.
The race for AI performance has intensified with AMD’s launch of its Strix Point Ryzen AI processors. These processors feature capable neural processing units (NPUs) integrated with discrete GPUs, offering a significant advantage in AI horsepower. Microsoft kicked off this battle with its requirement for a 40 TOPS (trillions of operations per second) NPU, but AMD’s new processors boast a 50 TOPS NPU and an additional 320 TOPS from the RTX 4070. This combination makes them the most powerful AI laptops on the market.
Intel’s Battlemage GPUs, the successor to the Arc Alchemist series, are expected to arrive in the second half of 2024 and bring significant performance improvements. With rumors suggesting a flagship model packing 56 Xe cores and a massive 112MB of L2 cache, Battlemage aims to compete in the midrange and budget segments of the graphics card market. Despite initial speculations targeting the RTX 4080, leaks now indicate that Battlemage is likely to take aim at the RTX 4070 instead. Intel is also rumored to be simplifying its lineup with Xe2, focusing on low-power applications and high-performance discrete GPUs.
