- About Us
-
Research Report
Semiconductors
LED
Consumer Electronics
Emerging Technologies
- Selected Topics
- Membership
- Price Trends
- Press Center
- News
- Events
- AI Agent
Cloud providers lift AI Capex to new highs; rack-scale GPU and custom ASICs rise, liquid cooling spreads.
The report highlights AI data center expansion by U.S. and Chinese CSPs. U.S. firms scale globally and invest more at home, while Chinese firms expand with self-developed chips, but both prioritize energy stability going forward.
Analyst summary: Samsung completed NVIDIA certification for HBM3e and began limited shipments; HBM3e will be the near‑term mainstream. Supplier competition is intensifying while HBM4 upgrades are expected to raise costs and premiums; certification timing will shape prices.
AI's surge is transforming data center storage. HDD faces high costs and long lead times due to new tech hurdles, creating a significant supply gap. This opens a key opportunity for QLC Nearline SSDs. NAND suppliers are accelerating advanced process transitions to dominate the AI-driven high-capacity storage market, positioning Nearline SSDs as the future mainstream.
TrendForce notes surging AI inference drives high-capacity storage demand. HDD shortage pushes CSPs to Nearline SSDs. NAND Flash vendors accelerate QLC validation & expansion. Future sees wider SSD adoption in AI, boosting enterprise SSD demand. NAND Flash expands to AI training, diversifying tech. Supply chain faces bottlenecks, requiring expansion & innovation.
Geopolitical constraints and import controls curb RTX PRO in China, with a special edition awaiting broader demand; CPX adopts GDDR7 memory with increased capacity, complementing Rubin GPUs that rely on HBMs, while near-term momentum remains with current solutions.
TrendForce indicates AI-driven data explosion causes structural changes in data center storage. HDD shortage accelerates QLC SSD demand, especially high-capacity products. Despite cost considerations, the widening HDD gap prompts North American CSPs to evaluate QLC SSD as cold data storage alternative. Supplier capacity constraints and CSP price sensitivity lead to short-term price negotiations, but in the long run, unresolved HDD issues will lead to explosive growth in QLC SSD demand.
AI compute faces memory bottlenecks. HBM's capacity/cost are limits. NAND Flash, via HBF (high-capacity near-memory) and AI SSD (intelligent preprocessing), optimizes AI architecture, redefining its value and role.
The iPhone 17 lineup officially welcomes a new ultra-thin flagship model. Apart from the base variant, all models in the series feature notable upgrades in design and hardware specifications. Although the overall smartphone market faces increasing challenges, both in demand and pricing, shipments are still expected to see slight growth over last year. The Pro series will remain the primary sales driver, while the market acceptance and sales performance of the new Air model will require time to be validated.
The surge in AI applications has fueled the development and growth of the semiconductor industry but has also brought new challenges. Apart from advanced process technology nearing physical limits and the increasing difficulty of process miniaturization, heat dissipation has emerged as a primary concern for major semiconductor foundries.
Chips used in AI servers, particularly GPUs, continue to improve in terms of performance. Consequently, the thermal design power (TDP) of individual chips has reached kilowatt levels. The traditional air cooling technology is approaching its physical limits, making way for the rise of liquid cooling technology. At the same time, the semiconductor industry and related academic research institutions are actively investing in the development of embedded liquid cooling solutions within chips.
