#micron-hbm-memory

#micron-hbm-memory

Samsung has reportedly swapped out its homemade OLED panels in some mid-range models for the Chinese CSOT-made OLEDs, aiming to reduce overall costs and avoid raising consumer prices.

After Google released its TurboQuant compression algorithm, which can purportedly drastically reduce the amount of memory required for certain AI workflows, stock prices for memory manufacturers dropped significantly.

With shares going for more than 38.0 times trailing price-to-earnings (P/E), I'm personally in no rush to chase the stock, especially with quarterly earnings just under a week away. That said, some big-name analysts have been bold enough to stay in the bull camp. And that's despite the hot, seemingly overheated run, the relatively stretched multiple, and uncertainties clouding the future of AI.

Zacks Investment Research assigned Micron its top Rank #1 (Strong Buy) rating this morning, citing the company's "significant role in the AI boom" and noting that Micron's 2026 HBM capacity is already sold out, driven by demand from AI chip partners including NVIDIA, Advanced Micro Devices, and Intel.

US PC shipments are set to fall by 13 percent this year thanks to the ongoing memory and storage crisis, with budget PCs hardest hit. Memory and storage costs will see at least a 60 percent increase during Q1 2026, compounding last year's rises of 40 to 70 percent.

Samsung and AMD share a commitment to advancing AI computing, and this agreement reflects the growing scope of our collaboration. From industry-leading HBM4 and next-generation memory architectures to cutting-edge foundry and advanced packaging, Samsung is uniquely positioned to deliver unrivaled turnkey capabilities that support AMD's evolving AI roadmap.

On paper, Positron's next-gen Asimov accelerators, no doubt named for the beloved science fiction author, don't look like much of a match for Nvidia's Rubin GPUs. Yet, the Arm-backed AI startup boasts its inference chip will churn out five times as many tokens per dollar while using one-fifth the power of Nvidia's latest accelerators to do it. Those are certainly some bold claims, which the company contends are possible because the chip was designed to support large-scale inference workloads.