🚀 Could #NVIDIA#'s next-gen chip specs change? Rubin Ultra's #HBM# configuration remains under review amid tight DRAM supply, while major CSPs weigh lower memory capacities.💡More analysis from #TrendForce#: 🔗

📈 #NVIDIA# RTX 50 graphics cards could rise by up to 30% in Korea from August as #TSMC# 4nm wafer and #GDDR7# memory costs increase. RTX 5090 (32GB) prices are already up by as much as KRW 1.5 million since mid-July.💡More: 🔗

📈 #Intel#'s #EMIB-T# is gaining momentum. #PSMC# reportedly secures exclusive supply of key silicon capacitors, while #UMC# sees strong silicon bridge demand. How far can the ecosystem expand? 💡More: 🔗

👀 AI servers can require up to 440,000 high-end #MLCCs# per cabinet, with suppliers including #Samsung# Electro-Mechanics pushing price hikes of up to 30%. Are passive components becoming the next supply chain bottleneck?💡More: 🔗

#HBF# is emerging as a complementary technology to #HBM# rather than a competing one, with the potential to address HBM's high cost and limited capacity. While HBM delivers unmatched memory bandwidth, it is difficult for it alone to meet the massive capacity requirements of #LLM# inference. HBF, on the other hand, leverages stacked Flash technology to provide sufficient bandwidth while offering several times the memory capacity of HBM at a significantly lower cost. Although #NVIDIA# has yet to announce any plans to adopt the technology, industry sentiment remains broadly positive. The interest shown by industry leaders such as #Google# and #Tenstorrent# have shown a willingness to consider adopting HBF, though specific use cases have not yet been determined. #TrendForce# believes the long-term architecture will likely combine HBM for ultra-high-speed computation with HBF for high-density data storage, creating a hybrid memory hierarchy that could become a key enabler of large-scale AI commercialization.

🚀 #Kioxia# is stepping up its AI #NAND# push, with BiCS 10 now in mass production and PCIe 6.0 and UFS 5.0 solutions set to follow later this year. Will this strengthen its position against #Samsung# and #SKhynix#? 💡More: 🔗

📢 AI computing demand remains strong. Capital spending by the world's nine largest CSPs is projected to rise about 90% YoY in 2026, surpassing US$886.7 billion.💡#TrendForce# on the latest AI server and CSP investment landscape: 🔗#Google# #Amazon# #Meta# #NVIDIA# #Microsoft# #Oracle# #ASIC#

📢 Taiwan's panel makers are reshaping their strategy. As #Innolux# and #AUO# divest aging factories, they're shifting from scale to high-value technologies such as Micro LED, AI, and CPO.💡More analysis from #TrendForce#: 🔗

💾 Server DRAM Isn't Safe Either: HBM's Crowd-Out Effect Is Not Easing. Cloud titans' aggressive AI infrastructure expansion continues to drive demand for HBM, server DRAM, and SOCAMM into 2027. With top suppliers prioritizing HBM in capacity allocation, the crowd-out effect has grown severe, leaving server DRAM supply increasingly constrained. TrendForce forecasts supply bit growth of just 24% in 2027, well short of market demand, sustaining a DRAM shortage throughout the year. Decode the 2027 DRAM deficit 👉 Explore the memory hierarchy shift 📖 #TrendForce# #SelectedTopics#

👀 China's chip equipment push is gaining momentum. #AMEC# targets 60% high-end equipment market coverage within five years, powered by an etching portfolio supporting 65nm-to-3nm manufacturing. 💡More: 🔗

👀 #MediaTek# confirms its AI ASIC collaboration with #Intel’s# EMIB-T advanced packaging. The chipmaker is now pushing 400G SerDes IP to target next-gen ASIC opportunities, including reportedly #Google’s# TPU v10 and Meta’s custom AI chips.💡More: 🔗

📈 #Samsung# Foundry’s revival is being unexpectedly fueled by memory. Strong demand for 4nm-based HBM4 base dies is lifting utilization, with Samsung reportedly aiming for 100% by year-end, up from 70–80% currently.💡More: 🔗

🔥 Fueled by orders from customers including NVIDIA and AMD, #TSMC#'s 3nm could reportedly reach 180K wafer starts/month 2–3 months ahead of market expectations, while 2nm remains on track to approach 100K by year-end.💡More: 🔗

👀 Zhongji Innolight, a leading supplier of #AI# optical transceivers, raised HKD 53.41B in Hong Kong's largest #IPO# of 2026, marking the city's biggest listing in seven years. 💡More: 🔗

👀 AI will continue to drive memory demand in 2027, keeping #DRAM# tight and prices firm. As #NAND# Flash capacity expands and consumer demand weakens, will second-half price corrections reshape the market? 💡More analysis #TrendForce#: 🔗

💡 Memory Is Buckling Under KV Cache Demand, Two Fixes Are in Play. Surging KV Cache demand against constrained memory supply created severe bottlenecks. Penguin Solutions, Marvell, and Meta are deploying CXL-based solutions to expand addressable memory capacity, while NVIDIA's KVTC and Google's TurboQuant compress demand from the other side. Get the full analysis 👉 Explore the memory bottleneck breakdown 📖 #TrendForce# #SelectedTopics#