Dan Nystedt

SK Hynix’s CEO said the global memory chip shortage is expected to last until the end of 2030 and that the risk of oversupply is low since memory chips are no longer “just commodities” in the AI era, media report. His comments came at a ground breaking ceremony for a US$4 billion advanced chip packaging plant in Indiana, USA, which aims to begin volume production of next-generation HBM (high bandwidth memory) chips in 2029. $SKHY $SSNLF $MU #semiconductors Links: https://t.co/XAPDwQG94t https://t.co/NFZ9Zo3p2j

Samsung Electronics aims to reach mass production on FOPLP (Fan-out Panel Level Packaging) by 2028, media report, adding its decades of work in display panels mean it’s taking a very different path than TSMC. Both are moving to rectangular shaped substrates – Samsung on 415 x 510 mm panel format, TSMC on 310 x 310 mm – but Samsung’s past work focused on mature chips. It’s expanding the effort to high-performance chips by 2028. TSMC, which calls its offering CoPoS (Chip-on-Panel-on-Substrate), is designed for high-end graphics chips (GPUs), custom ASICs, and other HPC (high-performance computing) chips. $TSM $SSNLF #Samsung #FOPLP #semiconduct…

Nvidia paused a deal with AI cloud companies that offered credit support for a share of revenue due to possible anti-trust concerns over the extent Nvidia may dictate how its customers do business, the WSJ reports, adding Nvidia may revamp the new business model, which it launched in July, or fold it into another program. The model is still in place for existing deals. https://t.co/oMMJ6cQ3y4 Also: https://t.co/QRwHHFcX3Z

Nvidia Key Comments From 2nd quarter earnings call Supply Chain Nvidia CFO Colette Kress: “Looking ahead, our preliminary expectation is for fiscal year 2028 revenue to grow approximately 70% year-over-year. Although we will work to close the supply-demand gap, we expect supply to remain a bottleneck at least through the end of fiscal year 28.” (Analysts expected 44% for fiscal 2028. Nvidia’s Fiscal 2028 starts Feb. 1, 2027, and ends Jan. 31, 2028) Nvidia CEO Jensen Huang: “Incredibly, we are seeing demand acceleration even at our scale. Customers’ forecasts point to our growth doubling next year. However, as mentioned earlier, we expect t…

Is DRAM worth its weight in Gold? Strong AI demand has driven memory chip prices up so fast that 1-kilogram of South Korea DRAM now costs as much as 620 grams of gold – up from 120 grams at the 2023 cyclical trough, according to South Korean trade data cited by media. That’s a 12.5x price surge in under 4-years. $MU $SSNLF $SKHY #semiconductors

Apple’s new M6 laptop/compute chip has continued the use of SoIC-MH advanced packaging, but high-end M-series chips (M6 Pro, Max, Ultra) may use 2 advanced packaging techniques, SoIC-MH for interconnect density and WMCM to integrate logic, LPDDR memory and high-speed I/O, media report, noting TSMC is readying capacity in 3 plants: Zhunan AP6, Longtan AP3 and Chiayi AP7. Analysts estimate WMCM capacity could reach 60,000 wafers-per-month (wpm) by end 2026 and over 120,000wpm in 2027. Important to note that the M6 is a monolithic die – everything on one slab of silicon – while the Pro, Max and Ultra versions are multi-die (chiplets) which need …

Kioxia plans to invest ¥1 trillion yen (US$6.3 billion) on a new fab in Japan’s Iwate Prefecture with operations to start in 2029, Nikkei Asia reports. The NAND flash memory maker is expected to receive some subsidies for the project. $SNDK $WDC $SSNLF $SKHY $AMAT $ASML $LRCX $KLAC #Kioxia #YMTC https://t.co/dhEMIzRJsz

Intel EMIB advanced packaging has become a confirmed rival to TSMC with strong yields and satisfied customers, media report, citing unnamed supply chain sources, who say Intel’s work with clients Google and MediaTek has been so smooth it indicates that costs and production yields are meeting targets. And with TSMC CoWoS capacity still tight, clients are likely to deepen work with Intel. $INTC $TSM $GOOGL #mediatek #semiconductors https://t.co/h4Bm35spoO EMIB: Embedded Multi-die Interconnect Bridge CoWoS: Chip-on-Wafer-on-Substrate

Nvidia Vera Rubin server rack data throughput is 30-times better than prior generation GB300 NVL 72, with cost-per-token dramatically lowered, giving Taiwan AI server manufacturers visibility to 2028, media report, citing unnamed industry sources. Nvidia announced performance data at the Hot Chips 2026 conference that just wrapped up. Taiwan’s Foxconn, Quanta and Wistron all expect to be in mass production of Vera Rubin servers in the 4th quarter. $NVDA https://t.co/0BeulOlEJ0

Nvidia’s Groq 3 LPX-based server racks have entered mass production at Foxconn, the primary supplier of the racks, media report, adding Nvidia announced the Groq 3 LPU chip in March. Samsung Foundry is producing the 4nm chips, which feature 500MB of on-chip SRAM per chip, better for inference (decode) and sidestepping the memory chip bottleneck. The Groq 3 LPX rack combines 256 LPUs for 128GB of total on-chip SRAM. 1/2 $NVDA $SSNLF #Foxconn #Samsung https://t.co/jCByuvICBa 2/2 Nvidia Groq 3 LPX servers require a different build due to rapid token generation and high frequencies, leading to some high-end material upgrades, media report: -High…

https://t.co/GEIVTlx7Oi Links: (Note: AMD already has a product page up for EPYC 9006 Series Server CPUs, but I still can’t find systems available to customers.) https://t.co/qJHYO6oug4 https://t.co/wnuKyJkXol https://t.co/dCt9LDcrxx It will probably be the last time for Apple. TSMC's 1.6nm (A16) will be in mass production by end-2026 and is designed for AI accelerators and other high performance computing chips via backside power delivery network (BSPDN). It's rumored Nvidia has reserved it all.

Japan will invest another ¥150 billion yen (US$940M) into chip foundry startup Rapidus, which aims to begin 2nm chip production in 2027, media report, adding the government has already invested ¥250 billion in Rapidus, and plans to provide ¥2.4 trillion in R&D aid. It is also considering another ¥300 billion in aid. In addition, three major Japanese banks will provide ¥2 trillion in loans to Rapidus, with the government guaranteeing 80% of the loans. $TSM $INTC $SSNLF #Rapidus https://t.co/5RiIvtyLWX

Xiaomi, the China smartphone giant, launched a new smartphone chip, the Xring O3, designed with Arm CPU and GPU cores and built by TSMC on 3nm, media report, adding the new chips mean lost sales for off-the-shelf chips like those made by Qualcomm and MediaTek. $ARM $TSM $QCOM #xiaomi #mediatek #semiconductors https://t.co/u0YuWALFnB

Nvidia’s top chip testing partner, KYEC, is expected to remain busy through year-end on Nvidia Rubin GPUs, which were postponed to the end of the 3rd quarter due to changes to the thermal design, media report, putting mass volumes in the 4th quarter. Strong demand for Nvidia Blackwell GPUs is keeping KYEC busy now. $NVDA #KYEC https://t.co/VqLvygRq2w

SK Hynix confirmed it will use Intel’s EMIB advanced packaging for next-gen HBM memory chips, joining Apple, MediaTek and other top TSMC clients in a win for Intel, media report, adding the victory is mainly due to the continued shortage of TSMC’s CoWoS capacity. $SKHY $INTC $TSM $AAPL #semiconductors https://t.co/nlfJmPxs2T

Rumor: Thailand and Vietnam could be hurt worst as big notebook PC brands turn back to China for mass production, media report, saying Southeast Asia turned out to be too costly for laptop production and US tariffs have not had as big an impact as feared. Taiwanese suppliers relocated production lines to Southeast Asia amid tariffs and client requests to move production outside of the Red Supply Chain (Communist nations (China)). But assembling a laptop in Southeast Asia costs US$9 more per unit than in China, which seems small but is most of the assembly cost. Pressure from local governments in China for Taiwan companies to make good on p…

Rumor: Taiwan’s Largan Precision is set to join Nvidia’s optical communications supply chain after completing end-to-end CPO (Co-Packaged Optics) module development, media report. -Mass Production: Expected 2nd-half 2027 (potential supply for Nvidia’s Rubin Ultra platform). -Components: FAs, MLAs, and integrated Fiber Array Units (FAUs). -Demand: Nvidia Spectrum-X shipments are projected at about 20,000 units this year, requiring 600,000–700,000 FAUs for onboard optical engines alone. $NVDA $TSM #Largan https://t.co/iTPwOcSMVQ https://t.co/ZaY0NJhMKr

Rumor: TSMC will spend NT$30 billion (US$940B) to buy two display panel plants from AUO and transform them into large-size advanced packaging plants (CoPoS and FOPLP), media report, adding the purchases are part of TSMC’s transformation of the Central Taiwan Science Park into a comprehensive ecosystem for Co-Packaged Optics (CPO), from 1.4nm (A14) fabs already under construction to these packaging plants. TSMC teams are doing on-site plant audits now, and if a deal can be reached soon, could have them cleared by end-1Q27. $TSM $NVDA #semiconductors #CPO #optics https://t.co/hj5zN4wZGb Apologies, US$940 million, not B.

Silicon wafer prices are set to rise 10%, the first major price hike for this key semiconductor material in over 3-years, media report, adding the hikes cover all sizes, 6-inch, 8-inch, 12-inch. End market demand has finally soaked up supply as the AI trend continues to draw in materials. Prices of specific products will continue to vary, but the overall impact on company revenues is expected to be average pricing up 10%. #SUMCO #ShinEtsu #GlobalWafers #WaferWorks #semiconductors https://t.co/edyxF5yXCj

Graphics cards makers are already mulling another round of price hikes after already raising 3rd quarter pricing amid tight supplies of GPUs and VRAM, media report. Prices for Nvidia RTX 5050 series and RTX 3060 series cards have already gained significantly this year, while the total available graphics card supply is seen contracting by at least another 20% this quarter. $NVDA $AMD #Asustek #Gigabyte #MSI https://t.co/dTmge9Kqoj

Taiwan’s ‘Big 5’ chip packaging and testing firms (OSATs) will spend over NT$460 billion (US$14.4B) this year on capex in a bid to catch up with AI-related demand, media report. Industry giant ASE said its biggest problem now is how to accelerate construction, equipment installation, and capacity rollout at 13 new projects and 8 renovations: ASE: NT$335 billion (US$10.5B) Powertech: NT$50 billion KYEC: NT$50 billion Sigurd Group: NT$20 billion ChipMOS: NT$7.2 billion in the 1st half. More coming $ASML $AMAT $TER $CAMT $NVMI #BESI #Advantest https://t.co/0j771DdatI

Nvidia CEO Jensen Huang touched down in Taiwan on Sunday for a brief visit, leaving after a few hours, media report, adding it’s his 3rd visit this year and comes just ahead of Nvidia’s earnings call this week. Media speculate the trip is personal as his itinerary was kept confidential, though also say there is much business to discuss with TSMC and other suppliers. He was photographed buying steamed buns in Taipei. $NVDA $TSM #semiconductors #AIservers https://t.co/kInqeHibok Also: https://t.co/eZMGBWucxU

Gigabyte Technology unveiled a desktop PC armed with the Nvidai GB300 Grace Blackwell Ultra Desktop Superchip, media report, a data-center-in-a-box with AI capabilities including allowing 400 people ask questions simultaneously. $NVDA https://t.co/EQkzWab8f8 Gigabyte’s W775-V10-L01: https://t.co/6Jm7EtL62O

Rumor: Robust AI demand has caused ABF supplies to tighten so far that key ABF film maker Ajinomoto has begun selective shipments, media report, adding some customers have been notified of a 30% reduction in supply. Taiwan’s, WaferChem Technology developed TBF (Taiwan Build-Up Film) as an alternative and may benefit from tight ABF supplies. $NVDA $AVGO https://t.co/dQtknQxN9p

Taiwan’s July export orders rose 61.9% year-on-year to a fresh record high US$97.94 billion and beating the expected 52.7% increase, media report. The US led all others with orders up 88.9% to $40.79B, the fastest pace of growth ever. Orders for IT/communications products, including servers & networking gear, rose 89.5% to $33.27B on strong AI and cloud service demand, while Electronics Products, including semiconductors and chip manufacturing services, rose 71.7% to $41.32 billion. In August, officials see export orders nearing $100B, up 61.4% - 64.8%. $TSM $UMC $ASX #Foxconn #Quanta #Wistron https://t.co/ZIujHpcB3W https://t.co/N9ue8XHhea

"OpenAI/Meta each will get 10% almost free AMD shares but Google will get 6.6% costly Marvell shares..."

Global Unichip, a custom ASIC chip designer that counts TSMC as its top shareholder, is raising NT$65 billion (US$2.04B) in funds to front risk production, testing & packaging, and material procurement for clients, media report. The fundraising includes an NT$40 billion syndicated loan and US$800M in convertible bonds. $AVGO $MRVL $GOOGL $AMZN $MSFT $META #semiconductors https://t.co/BxmTJPxKBl

Winbond, Taiwan’s 2nd biggest DRAM maker, has moved up its timeline to expand production in Kaohsiung, (south) Taiwan, media report, kicking off expansion at both phase 2 and phase 3 of the fab at the same time, with clean room construction to begin in 2027 and equipment installed by 2029. $SSNLF $SKHY $MU #Winbond #NanyaTech #semiconductors https://t.co/x7KMZ69JJp

https://t.co/dJnxYDBf7S

Cerebras, the maker of giant AI accelerator chips armed with SRAM onboard, announced its latest generation, the CS-4 is made on TSMC’s 5nm manufacturing process, media report. Cerebras says CS-4 servers are up to 30 times faster than servers that use GPUs. Cerebras chips do not use HBM memory, instead relying on SRAM. $CBRS $NVDA $AMD #semiconductors https://t.co/M9LkZcKfq6