This week’s hardware news covers AMD’s controversial RX 950 GPU launch criticized for its outdated specs and pricing, Nvidia’s employee detention amid GPU smuggling investigations, and memory market shifts with CXMT’s growth amid DRAM shortages. Additionally, major AI investments by Nvidia and AMD, concerns over data privacy in expanding AI data centers, and Noctua’s collaboration on 3D printer filament for custom cooling solutions were highlighted.
In this week’s hardware news recap, several notable developments were discussed, starting with AMD’s controversial launch of the RX 950 GPU. Priced at $280, this 50-class GPU features a 64-bit memory bus on its 4GB OEM model, a specification reminiscent of much older and less capable GPUs, which has led to criticism about its performance and value. The 8GB model includes a 128-bit bus but still requires an 8-pin power connector, which is unusual for a low-power card. The naming strategy and hardware choices have been called misleading and indicative of AMD “losing its mind,” as the card seems overpriced and underpowered compared to older models like the RX 580.
Another major story involves the ongoing GPU smuggling investigation in Taiwan, where an Nvidia employee was detained for alleged falsification of business documents related to smuggling activities. This comes amid Nvidia CEO Jensen Huang’s repeated public denials of any chip diversion or smuggling within the company. The investigation follows previous busts involving Super Micro and highlights the complex and lucrative black market for GPUs. Despite official denials, evidence suggests significant smuggling operations have taken place, raising questions about the integrity of supply chains and corporate transparency.
Memory supply issues continue to plague the industry, with Apacer warning of a potential 70% drop in DRAM availability by 2027, exacerbating already high prices and scarcity. This shortage is partly due to 60% of DRAM capacity being allocated to enterprise solutions, squeezing consumer availability. Meanwhile, Chinese memory maker CXMT has made significant strides, recently launching a successful IPO and capturing around 7.67% of the global DRAM market. Despite regulatory uncertainties surrounding its status in the US, CXMT’s rapid growth and potential partnerships with companies like Apple signal a shifting landscape in memory manufacturing.
In the AI and data center space, Nvidia is reportedly providing substantial financial backing—up to $250 billion—to OpenAI to support a massive data center project. This funding aims to secure infrastructure rather than directly supplying GPUs, illustrating the enormous scale and investment required for AI development. AMD is also engaging in this sector through a strategic partnership with Anthropic, committing up to $5 billion to deploy GPUs for AI workloads. However, concerns about data privacy and government surveillance are raised with the expansion of AI data centers globally, including a $650 billion project in South Korea aimed at securing AI capabilities faster than other nations.
Lastly, a lighter but interesting update was shared about Noctua collaborating with Pruscha to produce 3D printer filament matching Noctua’s iconic brown and beige fan colors in PETG material. This allows enthusiasts to create custom, heat-resistant fan shrouds for GPUs and other components, improving cooling and aesthetics. The filament is relatively expensive but offers better thermal stability than PLA, making it suitable for high-performance builds. The video wrapped up with reminders about restocked merchandise, upcoming content, and ways to support the channel.