StatementDog | AI & Tech Supply Chains
@statementdog
Tracking the companies and technologies behind the AI buildout. Research informed by company interviews, site visits, analysts, and industry experts.
92 Following 4.3K Followers
@LIWEI_TWCapital Agreed. The ABF prepayments you mentioned are a strong sign of customer commitment. With customers putting cash down years in advance, suppliers have a much firmer basis to expand.
Foxconn Interconnect Technology (FIT) is showcasing a 480kW liquid-cooled busbar system for power distribution within AI racks at the 2026 OCP Global Summit.
Its lineup also includes a 12.8T XPO platform that combines electrical and optical connections with direct liquid cooling, as well as a liquid-cooling design for switches with 32 OSFP ports.
The announcement does not disclose customer orders, shipment volumes or commercial deployment dates for these products.
Show more
C Sun’s Q3 revenue rose 7% from Q2 to NT$2.98 billion, but net profit attributable to shareholders fell 13.3% to NT$555 million, MoneyDJ reported on Oct. 14.
For the first nine months, the Taiwanese supplier of semiconductor and PCB manufacturing equipment posted revenue of NT$8.02 billion, up 84.2% year over year. Net profit rose 191.8% to NT$1.66 billion, or NT$10.67 a share.
The figures are preliminary. MoneyDJ attributed the sequential profit decline to product mix and non-operating items but did not quantify each factor’s impact.
Show more
Projects still in planning or preparation accounted for 71.6% of the total investment value in CINNO’s Q3 2026 survey of Chinese AI data centers.
The firm tracked 62 projects with a combined investment value of CNY 728.8 billion, up 34.4% year over year.
Projects at the delivery or production stage accounted for 13.6% of the total. Their combined investment value rose 30.5% year over year to CNY 99.36 billion.
These figures reflect project investment values by development stage, not cash spent during the quarter.
Show more
GlobalFoundries $GFS has been selected as the technology provider and operator for two planned fabs in Malta, New York, under a U.S.-Japan initiative with up to $21 billion in funding from Japan.
About $10 billion will go toward the first 300 mm fab, which is expected to come online in 2029. Once fully operational, it would add capacity for roughly 900,000 wafers a year. GF expects construction to begin later in 2026.
A further investment of about $10 billion would fund a second fab with similar capacity.
Show more
Taiwanese cooling supplier Auras unveiled a 6MW in-row coolant distribution unit (CDU) at the 2026 OCP Global Summit, MoneyDJ reported on Oct. 14.
The company says the system adjusts coolant flow and differential pressure as IT workloads change, using predictive controls to anticipate cooling needs. Its pumps, control units and filters feature a modular, tool-free design intended to simplify maintenance.
Auras is also extending liquid cooling to memory modules, PCIe interfaces and power systems. The report does not disclose customer orders or shipment volumes for the 6MW unit.
Show more
TrendForce now expects combined capital spending by nine major cloud providers to exceed $1.5 trillion in 2027, up 60% year over year. It previously forecast 50% growth.
The firm says higher server and component prices are driving up spending. Some suppliers also require advance payments to secure parts, bringing cash outlays forward.
It expects AI server shipments to rise nearly 36% in 2027, with revenue growing more than 90% to almost $920 billion, helped by higher average selling prices.
Show more
DDR4 and DDR5 module prices are rising in China’s channel market, while prices for some lower-capacity embedded storage products are falling, according to CFM’s Oct. 13 market update.
CFM attributes the DRAM price increases to tight chip supply and higher procurement costs. Some module suppliers are becoming more cautious about buying as PC makers resist further price hikes.
During the week covered by the report, prices fell for eMMC products with capacities of 64GB or less and for 64GB UFS products. CFM cites weak demand and high customer inventories, with some buyers reselling stock. These findings reflect conditions in the spot and channel markets.
Show more
Optical links could extend PCIe’s reach while preserving its architecture. Lumentum, Astera Labs and Corning are testing that approach in a rack-scale proof of concept.
At the 2026 #OCP# Global Summit, #Lumentum#, #AsteraLabs#, and #Corning# Optical Communications are demonstrating a rack-scale optical PCle link that combines:
🔹Astera Labs' Scorpio Smart Fabric Switch
🔹Lumentum's 1060 nm VCSEL-based optical interconnect solution
🔹Corning multimode fiber connectivity
The proof of concept uses multiple 32 Gb/s optical channels to demonstrate an error-free PCle-5 link— showing how optics can extend PCle connectivity within an Al rack while preserving a PCle-based architecture.
Show more
Semtech will invest to secure dedicated silicon germanium (SiGe) production capacity at Tower Semiconductor for its AI data-center interconnect chips.
The expanded multiyear agreement, announced on Oct. 13, covers Tower’s fabs in the US, Israel and Japan. It supports Semtech’s optical and copper interconnect products as data centers move from 800G and 1.6T connections toward 3.2T.
The companies did not disclose the investment amount, how much capacity is being reserved or how that capacity would be allocated across the three countries.
Show more
Panasonic Industry, a subsidiary of Panasonic Holdings (6752 JP), will triple prices for some of its general-purpose copper-clad laminates shipped from Jan. 1, 2027, Taiwan’s Commercial Times reports.
The increase applies to all R-1705 specifications and to listed products at least 1.1 mm thick in the R-1766, R-1566 and R-1755 series. It covers production at Panasonic’s plants in Japan, China, Taiwan and Thailand.
Panasonic said changes in market conditions and demand patterns had made current prices difficult to maintain for some products.
The company previously raised prices for its glass-epoxy multilayer copper-clad laminates by 30%, effective for shipments from May 1, 2026.
Kingboard Laminates (1888 HK) also raised prices by 15% for laminates and prepreg made with 2116-grade or thinner glass fabric, effective for new orders from Oct. 13. It was the company’s eighth increase of 2026. Kingboard cited tight supplies of glass fabric and yarn.
Show more
Marvell says its second-generation Alaska A 1.6T copper-link chip supports cable runs of more than six meters, twice the reach of its predecessor.
The digital signal processor (DSP) uses a 3nm process. Marvell lists its power consumption at 12.8 watts.
Marvell announced the chip alongside its new Alaska E 3.2T retimer on Oct. 13. Both devices are now sampling to customers for high-speed links within AI systems.
Show more
ASML is exploring ways to expand capacity in 2028 beyond the 30% increase it previously outlined for low-NA EUV and DUV immersion systems.
The Dutch lithography equipment maker said in its Oct. 14 earnings release that its planned supply would fall short of customer demand in both 2027 and 2028.
Q3 revenue came in at €11.66 billion, with a gross margin of 57.5%. ASML expects Q4 revenue of €14.2 billion to €15.2 billion.
Show more
CPO adoption will depend on how easily customers can monitor, cool and service the optical hardware. Wiwynn, Ayar Labs and GUC are testing those practical requirements in a working server.
$SIVE $NVDA $AMD
AYAR LABS' OPTICAL LINK IS NOW RUNNING INSIDE A CUSTOM ASIC PACKAGE, IN A WORKING SERVER, AT OCP. CO-PACKAGED OPTICS JUST WENT FROM RACK DISPLAY TO LIVE SYSTEM
SAN JOSE, Oct. 13 - Wiwynn (TWSE: 6669), Ayar Labs and Global Unichip (TWSE: 3443) are running their first live demonstration together at the OCP Global Summit: a Wiwynn compute system carrying GUC's ASIC with Ayar's TeraPHY optical engines packaged alongside it, lit by Ayar's SuperNova external light source and streaming the engines' health data in real time. The three companies call it a step up from their earlier rack-level architecture work to "platform integration and functional validation."
- GUC multi-chip packages with TeraPHY optical engines inside, fed by ELSFP-based high-power SuperNova light sources, in an ORV3-ready compute system
- Telemetry polled continuously from the engines inside the packages, "proving the optical link is up inside the system"
- As the tuning loop tracks thermal and wavelength drift, eye height, OMA and extinction ratio hold steady on screen
- Detachable fiber array unit and in-chassis fiber routing, built for servicing and manufacturing
- Wiwynn's own cooling design keeps optical power stable under real operating conditions
"The bar is no longer whether it works; it's whether it can be deployed and managed at scale," Ayar CEO Mark Wade said in the release.
Ayar counts NVIDIA ($NVDA) and AMD ($AMD) among its strategic investors, so both GPU camps already own a piece of the engine being validated here. The light source is SuperNova, and Sivers ($SIVE) has been part of it since 2022: its DFB laser arrays lit SuperNova at ECOC that year, it booked a $1 million Ayar order for light source technology in July 2023, and the two expanded the partnership toward high-volume manufacturing in December 2024.
Show more
Sigurd Microelectronics, a Taiwanese chip-testing company, says a subsidiary has agreed to buy a factory building and related equipment from CyberTAN for NT$2.19 billion to expand its advanced-testing operations.
The company plans to use the Hsinchu Science Park site to test AI ASICs and CoWoS-packaged chips, as well as silicon-photonics and high-speed optical-communications chips, Anue reports.
Sigurd estimates the site could add about 25% to its current testing capacity once volume production begins. The report does not specify a production start date for the acquired site.
Show more
SEMI forecasts global spending on equipment for 300mm front-end fabs will rise 31% year over year to $149 billion in 2026 and reach $227 billion in 2029.
Its Oct. 13 outlook also projects that construction will begin on a combined 38 new fabs or production lines worldwide across 2026 and 2027, excluding facilities dedicated solely to R&D.
Separately, SEMI expects fab construction spending to rise from $47.5 billion in 2026 to $68.4 billion in 2028. The forecasts point to a multiyear expansion in manufacturing capacity, but they do not represent orders already booked by equipment suppliers.
Show more
$AMAT is targeting share gains and higher non-GAAP margins by 2030, with advanced logic, DRAM, packaging and services driving the plan.
We've presented our longer-term growth expectations and outlined how our leadership in materials engineering positions the company to outperform the wafer fab equipment market over the coming years.
Learn more:
Show more
KYEC, a Taiwanese chip-testing company, says its board has approved a prepayment agreement with Nvidia to support plans to build testing capacity in the US.
Under the agreement, Nvidia will make advance payments, while KYEC and its wholly owned US subsidiary, AeternZ Semiconductor Testing, will add testing capacity in the US and reserve it for Nvidia. The payments will be credited against future testing fees.
The Oct. 13 filing did not disclose the prepayment amount, how much capacity the agreement covers or when that capacity would be available.
Show more
SK hynix has unveiled an SSD controller design that distributes routine storage tasks across about 100 small H-TPU processors, The Elec reports, citing an Oct. 12 presentation.
A general-purpose CPU handles more complex work, such as error recovery. SK hynix plans to use the design in its next-generation enterprise SSD controllers.
The company says each H-TPU uses about 1/40 as much power and occupies 1/25 as much chip area as a general-purpose CPU core. Those figures compare individual processors, not complete SSDs.
Show more
AMD’s new Helion rack architecture is designed to let cloud operators dedicate CPU resources to running AI agents while handling model inference on separate systems.
The CPU-only configuration supports up to 32,768 EPYC cores per rack. An all-in-one version combines EPYC CPUs, Instinct MI350P GPUs and general-purpose compute in a single rack.
AMD says it is developing Helion configurations with ecosystem partners and plans to make them available in November 2026. Foxconn, QCT and Wiwynn are among the companies supporting the design.
Show more
