Semiconductor News & Analysis Feed
128 articles
2026-06-18
wccftech.com
2026-06-18
Wccftech
SK Hynix has commenced sampling of its next-gen HBM4E memory, offering up to 16 Gbps speeds and 48 GB capacities.
The acceleration in AI has pushed DRAM manufacturers to dial up their development plans. As such, SK Hynix is racing its rival Samsung to deliver the first HBM4E memory solutions to partners who will be using it to power their next-generation datacenters.
The HBM4E memory will play a
2026-06-18
wccftech.com
2026-06-18
Wccftech
By now, everyone and their grandmother is familiar with ASML, the sole provider of EUV (Extreme Ultraviolet) capable lithography machines in the world.
As hyperscalers continue to aggressively build out compute, we constantly discover new bottlenecks at almost every level of Jensen's five-layer AI cake, down to an insulating film supplied by, of all things, an MSG manufacturer. Yet for all the no
2026-06-17
wccftech.com
2026-06-17
Wccftech
The Taiwan Semiconductor Manufacturing Company (TSMC) is expanding its supply chain for glass substrates as part of the first step of establishing a supply chain for the chip-on-panel-on-substrate (CoPoS) packaging technology, according to supply chain sources. The glass substrates will be used in the firm's advanced versions of the chip-on-wafer-on-substrate (CoWoS) packaging technology, and TSMC
2026-06-16
wccftech.com
2026-06-16
Wccftech
The U.S. trade sanctions on Huawei pretty much mean the latter is fighting an uphill battle with other chipmakers, as it lacks access to the latest and greatest in EUV equipment. Fortunately, a new report states that later this year, when the Mate 90 series materializes, the former Chinese giant will finally show to the world that its use of DUV machinery is sufficient to mass produce competitive
2026-06-16
wccftech.com
2026-06-16
Wccftech
NVIDIA is releasing RTX Remix 1.5 today, which brings with it a range of new features, including support for new RTX IO compression.
This year, NVIDIA's RTX Remix suite is getting some major updates. We've already seen the addition of RTX Remix Logic with Dynamic FX and new Particle VFX. Today, NVIDIA releases RTX Remix 1.5, which is going to take things a step further.
With RTX Remix 1.5, NVIDI
2026-06-16
wccftech.com
2026-06-16
Wccftech
Memory makers are now adopting domestically produced larger size DDR5 memory chips for their DDR5 modules.
Instead of depending on the three memory giants, such as Samsung, Micron, and SK Hynix, some memory vendors in China are now adopting domestically produced DDR5 chips. As per IT Home, two vendors known as Gloway and KingBank have introduced their latest DDR5 memory modules, using the 24 Gb m
2026-06-16
wccftech.com
2026-06-16
Wccftech
The memory requirements of experiencing the entire Siri AI features suite will eventually force Apple to introduce a minimum of 12GB LPDDR5X RAM for its iPhone 18 lineup, but more importantly, is how increased demand for these devices will contribute to the existing memory crisis while making a handful of suppliers wealthy in the process.
2026-06-16
wccftech.com
2026-06-16
Wccftech
Memory manufacturer Micron could benefit from high DRAM and HBM prices in 2026 and 2027, believes financial firm Aletheia Capital. Micron, like SK hynix and Samsung, has benefited from the strong demand for AI chips in the current buildout as memory production constraints have pushed prices up. The financial firm believes that the value of memory chips in the AI ecosystem will become the most impo
2026-06-16
wccftech.com
2026-06-16
Wccftech
US-based AI company, Tensordyne, has announced the successful tape-out of its Napier chip, which it claims to demolish NVIDIA's Blackwell & Rubin chips with leading token throughput and efficiency.
The Napier chip will be the core component of the Tensordyne Napier TDN system, which is designed in collaboration with Broadcom and HPE Juniper Networks. The Napier platform has one goal: to unify AI
2026-06-16
wccftech.com
2026-06-16
Wccftech
The compact space inside smartphones makes it difficult to adopt High Bandwidth Memory (HBM), not to mention the thermal problem, but a rumor claims that Chinese smartphone brands have a custom-made Low Latency Wide DRAM (LLW) that adopts a similar integrated design as the aforementioned technology. While the design isn’t “true HBM,” it aims to resolve the performance bottleneck issue that LPDDR R
2026-06-16
wccftech.com
2026-06-16
Wccftech
The datacenter segment is preparing for the first wave of next-gen 800V HDVC systems that will power NVIDIA & Google infrastructure this year.
As Datacenters continue to outstrip electricity generation, the need for more power-efficient & versatile solutions has become essential. This is why major companies like NVIDIA & Google have invested heavily in 800V DC or HVDC (High-Voltage Direct Current
2026-06-15
wccftech.com
2026-06-15
Wccftech
TSMC is expected to raise prices of its 2nm wafers, which might lead NVIDIA and Apple to consider using Samsung as an alternative.
We know that TSMC is already ramping up volume production of its cutting-edge 2nm process, which is also one of the most advanced technologies in the world. But the latest tech also comes at a high price. This is due to significantly higher demand & markups driven by
2026-06-15
wccftech.com
2026-06-15
Wccftech
The 2nm GAA process is slowly gaining traction through improved yields, with Samsung’s more mature 4nm and 8nm nodes increasing in utilization, becoming one of the reasons why the Korean giant is slowly approaching the path of profitability. Unfortunately, there’s always some resistance in the way, and according to the company’s semiconductor head, it’s performance-based bonuses and a mobile-centr
2026-06-14
wccftech.com
2026-06-14
Wccftech
NVIDIA's Blackwell GB300 has posted record performance in AA-AgentPerf, a new benchmark that measures Agentic AI workflows.
Artificial Analysis has a new benchmark out called AA-AgentPerf, which measures how many active agents an inference deployment can support under realistic workloads, which include:
The AA-AgentPerf benchmark is used to measure three key metrics, which form the basis of mode
2026-06-13
wccftech.com
2026-06-13
Wccftech
NVIDIA's fastest PRO graphics card, the RTX 6000 Blackwell, continues to see price hikes with the latest official listing of $13,250.
Last month, we reported that NVIDIA's RTX PRO 6000 Blackwell GPUs had breached the $10,000 US price point, but that was just the beginning. Now, NVIDIA has officially listed the card for an even higher price.
Over at the official NVIDIA marketplace, the NVIDIA RTX
2026-06-13
wccftech.com
2026-06-13
Wccftech
NVIDIA's fastest PRO graphics card, the RTX 6000 Blackwell, continues to see price hikes with the latest official listing of $13,250.
Last month, we reported that NVIDIA's RTX PRO 6000 Blackwell GPUs had breached the $10,000 US price point, but that was just the beginning. Now, NVIDIA has officially listed the card for an even higher price.
Over at the official NVIDIA marketplace, the NVIDIA RTX
2026-06-13
wccftech.com
2026-06-13
Wccftech
AMD has officially launched its Ryzen AI Halo AI PC, which offers super-fast token throughput at a $3999 price point.
The first AMD Ryzen AI Halo platforms are now available on retail, and are listed at the official MSRP of $3999. These systems pack the Ryzen AI MAX+ 395 SoC, 128 GB of memory, and lots of AI horsepower within a compact form factor that challenges NVIDIA's DGX Spark at a much lowe
2026-06-12
wccftech.com
2026-06-12
Wccftech
NVIDIA has reportedly reached out to several Chinese clients to adopt its upcoming Vera CPUs & there is a lot of interest building up.
NVIDIA's GPUs have faced a hard time in China. The company saw US regulations restrict them from selling Hopper chips, while the more advanced Blackwell generation is entirely banned. NVIDIA's CEO stated a few months back that their China share has dropped to Zero
2026-06-11
wccftech.com
2026-06-11
Wccftech
SK Hynix is expected to begin mass production of its 375-Layer NAND Flash solution by the end of 2026, offering higher storage capacities.
SK Hynix and Samsung are in the race to reach the highest number of layers in the NAND Flash segment. While Samsung is aiming beyond 400-layers with dual stacking solutions going up to 1000-layers, Hynix will soon be commencing mass production on its 375-Layer
2026-06-11
wccftech.com
2026-06-11
Wccftech
With multiple supply chain reports focusing on Intel's EMIB-T chip packaging technology, analyst Ming-Chi Kuo has shared that TSMC's next-generation packaging technology, CoPoS, will enter mass production in 2028. CoPoS, short for chip-on-panel-on-substrate, seeks to overcome the limitations of the current CoWoS (chip-on-wafer-on-substrate) packaging technology by increasing the area on which the