TLDR: AMD has reportedly increased the price of its Instinct MI350 AI GPU by 70%, from $15,000 to $25,000. This strategic move aims to position the MI350 as a direct competitor to NVIDIA’s Blackwell B200, leveraging its enhanced performance and memory capabilities for AI workloads. The price hike reflects strong demand for AMD’s AI products and is expected to positively impact the company’s data center revenue.
Advanced Micro Devices (AMD) has made a significant strategic move in the highly competitive artificial intelligence (AI) chip market, reportedly increasing the price of its Instinct MI350 AI GPU by a substantial 70%. The price per unit has risen from an estimated $15,000 to $25,000. This bold adjustment, reported on July 28, 2025, signals AMD’s confidence in the MI350’s performance and its aggressive intent to challenge NVIDIA’s dominance in the AI accelerator space.
Despite the considerable price hike, the MI350 remains more affordable than NVIDIA’s entry-level Blackwell B200, which is expected to retail between $30,000 and $40,000. AMD’s decision is rooted in the belief that the MI350’s enhanced compute and bandwidth capabilities now directly rival NVIDIA’s offerings, justifying the premium.
The Instinct MI350, part of the MI350 series which includes the MI350X and the flagship MI355X launched in June, is designed for demanding AI training and inference workloads. It boasts 288GB of HBM3 memory, which is 60% more than NVIDIA’s B200, and offers twice the 64-bit floating-point performance. This makes the MI350 capable of handling AI models with up to 520 billion parameters on a single node, a critical advantage for enterprises working with large language models (LLMs) and complex simulations. The MI350 series is built on TSMC’s 3nm process with CDNA 4 architecture, featuring up to 185 billion transistors and advanced cooling. It delivers impressive peak performance, reaching 20 PFLOPS, a quadruple increase over the previous MI300X, and a 35-fold improvement in inference performance.
This price increase indicates robust demand for AMD’s AI products and suggests a potential boost in the company’s quarterly revenue. Morgan Stanley analysts project that the new pricing could add approximately $500 million to AMD’s data center revenue in the next quarter if customer adoption remains strong. HSBC analysts have expressed optimism regarding AMD’s competitive edge, raising their target price for AMD stock to $200 and forecasting AI chip revenue to reach $15.1 billion by 2026, a significant increase from previous estimates of $9.6 billion.
Investors have reacted positively to the news, with AMD shares jumping around 4% on Monday, July 28, 2025. This surge reflects investor confidence in AMD’s pricing power in the burgeoning AI chip market and the potential for increased margins and top-line growth. Wells Fargo analysts have also reiterated their positive outlook, increasing their estimates for AMD’s Data Center GPU sales. The focus for AMD’s upcoming fiscal second-quarter results on August 5 will be the ramp-up and market penetration of the MI355X data center GPUs, which began shipments in June.
Also Read:
- ASUS Unveils ExpertCenter Pro ET900N E3: A Desktop Supercomputer with NVIDIA GB300 Superchip Delivering 20 PFLOPS AI Performance
- Enterprise Software Sector Witnesses AI-Driven Growth and Strategic Investments
While AMD’s hardware advancements are compelling, the broader ecosystem and software integration, particularly NVIDIA’s established CUDA platform, remain key factors in the AI chip race. However, AMD’s aggressive pricing and performance strategy, coupled with its product roadmap including the upcoming MI400 series (2026) with HBM4 technology, underscore its ambition to compete at the system level with offerings like the Helios rack-scale system, directly targeting NVIDIA’s GB200 NVL72.
