DELIVERCHIP（HK）LIMITED — 30 March: This Week’s News

Highlights of the Week

1. Intel and AMD to raise prices across their entire product ranges from April

2. Baiwei Storage secures a major $1.5 billion wafer procurement contract

3. Google unveils a new, cutting-edge AI memory compression algorithm

Intel and AMD to Raise Prices Across Their Entire Product Ranges from April

According to supply chain sources, Intel and AMD have separately notified their customers that they will be raising prices across their entire range of processors from April. At the same time, delivery lead times for the relevant products have lengthened significantly—from the normal level of a few weeks to several months or even longer.

Price pressure in the PC hardware market is continuing to spread from the storage and memory sectors to the core processor segment. The CPU market has currently seen a round of price increases averaging approximately 10 per cent to 15 per cent, affecting both server chips and consumer-grade products, marking the latest sign of widespread hardware cost inflation.

Baiwei Storage Signs US$1.5 Billion Wafer Procurement Contract

On the evening of 24 March 2026, Baiwei Storage issued an announcement stating that it had signed a routine operational procurement contract with a memory original equipment manufacturer (OEM), with a total committed purchase value of US$1.5 billion (approximately RMB 10.3 billion), securing a two-year supply.

The contract aims to secure the supply of memory wafers for the next two years in advance, enhance the stability of the company’s medium- to long-term memory wafer supply, and mitigate the impact of price fluctuations on costs. This will help the company safeguard its raw material supply and stabilise production costs against the backdrop of a continuously improving market outlook in the memory industry.

Google Unveils New Extreme AI Memory Compression Algorithm

On 25 March, Google Research unveiled a new extreme compression algorithm called TurboQuant, claiming it can compress the ‘key-value cache’ (KV Cache)—the most memory-intensive component in the inference process of large AI models—by at least six times without any loss of precision, whilst achieving up to an eight-fold performance boost on NVIDIA H100 GPUs.

Following the announcement, a chain reaction ensued – the global memory chip sector saw a collective plunge, with share prices of companies such as SanDisk, Micron Technology and SK Hynix falling sharply, reflecting shifting market expectations regarding AI memory demand.

All news in this article is sourced from the internet.