Latest News - Page 424

ASUS is cooking up its new ROG NUC pre-built PCs for CES 2025, with a new tease that the next-gen Mini-PC will feature Intel's new Core Ultra 200H "Arrow Lake-H" CPUs and NVIDIA's new GeForce RTX 50 series "Blackwell" GPUs.

In a new leak on Chinese social media platform Weibo, we get some sneaky photos of the next-gen ASUS ROG NUC system, which from the outside doesn't look all that different. ASUS is baking in all of the changes inside of the new ROG NUC, where we'll see the use of Intel's new Core Ultra 200H series CPU, and NVIDIA's next-gen GeForce RTX 50 series GPU.

We should see ASUS using the flagship Core Ultrra 9 285H processor, as the current-gen ROG NUC (more details on that in the link below) features the flagship Core Ultra 9 185H "Meteor Lake" processor. The new ROG NUC will feature an RTX 50 series GPU -- expect to see the RTX 5070 or RTX 5060 -- while the current-gen ROG NUC rolled out with up to the RTX 4070 inside.

Samsung has failed to secure another large customer for its semiconductor business, losing Qualcomm and its contract to build its next-gen Snapdragon 8 Elite 2 processor, which will now be fabbed on TSMC's new N3P process node.

In a new article from The Bell, we're learning that this is yet another painful experience for Samsung, as it hasn't proven its semiconductor business can mass produce high-end chips. According to industry insiders, Qualcomm's new Snapdragon 8 Elite 2 (tentative name) will be released in the second half of 2025, mass-produced on TSMC's third-gen 3nm (N3P) process node. Samsung entered the competition to win the order, but "ultimately fell through".

Samsung used to be the largest manufacturer of Qualcomm chips, with a monopoly until the company launched its first-gen Snapdragon 8 processor in late 2021. The company experienced yield issues, but TSMC was then put in charge of the improved version -- Snapdragon 8+ -- and the rest is history.

Intel has confirmed its CES 2025 press conference, where we can expect to see some "groundbreaking innovations" for both the client computing and AI market.

Intel will kick off its CES 2025 press conference before AMD and NVIDIA, starting on January 6 at 8:30AM PST, with the company describing it as "Intel-Driven Keynote: AI Inside for a New Era". This will be the very first time that we get to see Intel's new co-CEO Michelle Johnston Holthaus, alongside SvP of Client Computing Group, Jim Johnson, and the first time in many years without the now-gone Pat Gelsinger.

We should expect a ton about AI (duh) but we'll also see Intel launch its new Core Ultra 200HX "Arrow Lake-HX" laptop processors, as well as the non-K versions of the desktop-bound Core Ultra 200S series "Arrow Lake-S" processors. Intel is also expected to show off some budget motherboards based on the new LGA 1851 socket, which should materialize into the B860 series.

SK hynix has begun preparations for the mass production of its new 16-Hi HBM3E memory, weeks after unveiling the world's first 48GB 16-Hi HBM3E memory at the SK AI Summit in Seoul.

In a new report from ETNews, we're learning that SK hynix is now integrating new equipment for 16-Hi HBM3E memory production, and optimizing its existing facilities. Production tests have reportedly already begun, with supply of the new 16-Hi HBM3E memory expected to flow in 1H 2025.

An industry insider told ETNews: "It is the initial line preparation process for 16-layer HBM3E mass production. I understand that the results of the major process tests are also coming out well".

AMD's flagship Ryzen 9 9950X3D processor will make its grand appearance at CES 2025, but new leaks tease the 16-core, 32-thread X3D beast on CPU-Z, check it out:

The upcoming Ryzen 9 9950X3D processor will have the same clock speeds as their non-X3D counterparts, but with some additional stacks of L3 cache through AMD's in-house 3D V-Cache technology. The flagship AMD Ryzen 9 9950X3D processor will feature 16 cores and 32 threads of Zen 5 power, with CPU clocks of up to 5.65GHz (5650MHz) in its engineering sample (ES) form.

AMD's new powerhouse Ryzen 9 9950X3D will also retain the same 170W TDP rating that the Ryzen 9 9950X has, but with much more case: 96MB + 32MB for a total of 128MB of L3 cache. This is an additional 64MB of L3 cache for the 9950X3D over the 9950X, so it'll be interesting to see the adoption of this new processor (gamers will want to buy the 9800X3D).

Clevo is cooking up its next-gen X580 gaming laptop, which is "just about to be launched" with an 18-inch display, Arrow Lake-HX processor, RTX 50 series Laptop GPU, and more.

The new Clevo X580 gaming laptop was teased in new leaks with the full spec sheet reveals we can expect the 18-inch laptop to feature an Intel Core Ultra 200HX series "Arrow Lake-HX" processor, up to 192GB of RAM, and NVIDIA's new GeForce RTX 50 series Laptop GPUs.

Clevo will be using the new Arrow Lake-HX processors with a 55W TDP, meaning we'll see chips with up to 24C/24T (no Hyper-Threading on Arrow Lake). Clevo's new X580 gaming laptop will feature the "NVIDIA GN22 Board 1" GPU, which might point to the soldered GPU on the laptop motherboard, so we could expect to see the RTX 5090 + RTX 5080 Laptop GPU, or an RTX 5070 Ti + RTX 5070 Laptop GPU inside of the Clevo X580 gaming laptop.

Intel's new high-end Arrow Lake-HX laptop processors aren't far away, with the new Core Ultra 9 275HX processor spotted on Geekbench, check it out:

The new Core Ultra 9 275HX processor was spotted inside of a new Razer Blade 18 gaming laptop with 24 cores and 24 threads, boosting at up to 5.3GHz and sporting 36MB of L3 cache. The Core Ultra 9 275HX will be one of the higher-end models of the new Core Ultra 200HX family, with the flagship Core Ultra 9 285HX taking the top spot.

Intel has dropped Hyper-Threading (HT) from its Arrow Lake CPUs in both desktop and laptop form, with the Core Ultra 9 275HX featuring a lower boost clock than the 285HX, with the 275HX featuring a base clock of 2.7GHz and boost clock of up to 5.3GHz with a maximum boost of 5.5GHz. We should also expect an Xe-LPG integrated GPU with 64 EUs and a boost clock of up to 2.0GHz.

TSMC is the king of the technology industry, controlling a dominant 90% of the world's advanced semiconductor manufacturing... and it's leaving South Korea shaking in its semiconductor boots.

In the rise of AI, South Korea's Samsung has been continuously falling behind -- especially with fellow South Korean memory rival SK hynix -- and that Korean media are now pointing out the fact that the South Korean government is well aware of the problem, and has plans to kick start a Korean Semiconductor Manufacturing Company, or KSMC.

KSMC would have the aim of following TMSC's successful model and would support the development of the local semiconductor industry, with experts saying that KSMC would face obvious challenges including the large technological gap with its competitors, and the lack of an industrial supply chain.

ASML is selling a Lego set of its bleeding-edge TWINSCAN EXE:5000 High-NA EUV lithography machines, but it's unfortunately only available to ASML employees.

There are many on social media calling for ASML to make its Lego set available to people outside of the company -- myself included -- but ASML spokesperson Monique Mols responded saying "sorry, but it's only for ASML employees".

Mopls was pressed about it again, but Mols said that the Lego model of the TWINSCAN EXE:5000 High-NA EUV lithography machine in its awesome Lego form is only a "fun iten for employees" as the company is celebrating its 40th anniversary. The Lego-style replica was selling for $216 on ASML's employee website, but the real machine costs over $350 million and weighs as much as two Airbus A320s... you know, the planes.

AMD's next-gen Radeon RX 9070 XT graphics card rumors are swirling, with the flagship RDNA 4 chip said to boost at up to 3.1GHz and have a 260W to 330W TBP depending on the AIB model.

In a new post on Chiphell, we're learning that the flagship Radeon RX 9070 XT graphics card will have a reference board design with a TBP (Total Board Power) of around 260W, while custom Radeon RX 9070 XT graphics cards will chow down on up to 330W of power. This is a 27% difference, allowing for some heavy overclocking on the better-built custom 9070 XT graphics cards.

AMD's new Radeon RX 9070 XT is expected to have a 2.8GHz base GPU clock, with boost clocks of the flagship RDNA 4 chip hitting 3.0GHz to 3.1GHz depending on the model.

NVIDIA and AMD are reportedly shipping their new GeForce RTX 50 series and Radeon RX 9000 series GPUs from China to warehouses in the US before President Trump's new tariffs kick in on January 20, 2025.

Both companies are expected to unveil their new Blackwell and RDNA 4 gaming GPUs at CES 2025 in early 2025, with NVIDIA expected to release 6 new GeForce RTX 50 series GPUs across Q1 2025, with new graphics cards launched in the RTX 50 series GPU family between January and March, lining up with previous rumors.

In a new report from Ctee, we're hearing that in response to an unclear implementation schedule of the new tariff increase after President Trump takes office on January 20, 2025, graphics card supply chain industry partners have said that both major GPU manufacturers (NVIDIA and AMD) have asked their partners to ship custom AIB cards produced in mainland factories in China in advance, before January 20, to warehouses in the United States.

We've got another look at the purported PCB of NVIDIA's next-generation GeForce RTX 5090, showing off the massive GB202 GPU and this time, we see Samsung 28Gbps GDDR7 memory modules. Check it out:

In a Christmas Day leak, we saw the PCB of a custom GeForce RTX 5090 graphics card in a compact, but crowded PCB that looks insanely powerful with its array of components, but now we've got a closer look at the PCB and those juicy GDDR7 memory modules.

NVIDIA's new GeForce RTX 5090 PCB leak is coming from the Baidu forums, with the PCB fully decked out for the holidays with the GPU, VRAM, and VRMs. The GPU is labeled as the GB202-300-A1 GPU which lines up with previous leaks, with this most likely being a qualification sample that should be close to (but not the) final design.

AYANEO has just announced the "world's first modular handheld" with the introduction of the new AYANEO 3, which has the ability to swap out modular buttons and controller layouts, to personalize your handheld in a totally new way. Check it out:

The new AYANEO 3 features the company's new technology "Magic Module" which has been in development for a couple of years, with button types on gaming handhelds no longer determining if the handheld is suitable for a particular game, or game style. Now, that's no worries as AYANEO 3 has modules that you can swap in and out, changing up the buttons and controller layouts for full personalization of your gaming handheld.

AYANEO 3 has multiple modules, touchpads, and 6 buttons to choose from, which look to be held in with magnets. The company has confirmed that the modular buttons can even be inverted, which is a wicked detail to see. This means that if gamers want the D-pad lowered, it takes just a couple of seconds to swap it out. Not only that, but gamers can adjust individual ABXY buttons if their placements cause text to appear in the wrong orientation, another nice touch.

Take-Two CEO Strauss Zelnick has provided some more GTA 6 hype over Christmas, saying in a recent interview: "I would say I think it'll be extraordinary and worth waiting for".

Zelnick was in a video interview with Conner Mather when he provided the new commentary on Grand Theft Auto 6, but they later talked about GTA's community feedback, and that Rockstar looks to have prepared features and content for the fans that they might not have even dreamed of... exciting stuff.

Zelnick continued: "We definitely listen to feedback from our fans, absolutely. You need to, in our business, and you're able to. But remember, the biggest hits are always the ones that are unexpected. Everything that we do has to be new and different, not derivative. So, by definition when people are commenting on something that is in the market, that means they're looking at something that was done before".

HP's new EliteBook X G1a laptop has leaked, with the new 14-inch workstation laptop rumored to feature AMD's flagship Ryzen AI MAX+ PRO 395 "Strix Halo"A APU.

The new AMD Ryzen AI Max+ Pro processor -- apart from its ridiculous name -- is the flagship Strix Halo APU, featuring 16 cores and 32 threads of Zen 5 goodness with a base CPU clock of 2.90GHz. The new HP EliteBook X G1a will use the flagship Strix Halo APU alongside support for up to 128GB of RAM.

HP will ship the new 14-inch EliteBook X G1a laptop in two different CPU configurations: the AMD Ryzen AI Max+ 395 (16C/32T) and the Ryzen AI Max 390 (12C/24T). AMD's new Strix Halo APU ships with a far more powerful integrated RDNA 3.5 GPU solution, with 40 Compute Units (versus 16 CUs on Strix Point) in its flagship 16-core, 32-thread form, with 32 CUs for the 12-core, 24-core variant.

Apple's next-generation M5 series processors are in the headlines, with rumors swirling that they'll use TSMC's advanced N3P process node, and server-grade SoIC advanced packaging.

In a new post to X by insider Ming-Chi Kuo, we're learning that Apple's next-gen M5 series processors entered the prototyping phase "a few months ago" with the M5, M5 Pro, M5 Max, and M5 Ultra processors entering mass production in 1H25, 2H25, and 2026, respectively.

Apple's new M5 Pro, M5 Max, and M5 Ultra chips will reportedly use server-grade SoIC packaging, with the iPhone giant to use 2.5D packaging called SoIC-mH (molding horizontal) to improve production yields and thermal performance, with separate CPU and GPU designs.

SK hynix is delivering a huge HBM memory order to Broadcom in 2025, which it will use on its new AI chips that it is making for cloud service providers (CSPs) including Google, Meta, and TikTok parent company ByteDance.

SK hynix already has NVIDIA as a major client of its industry-leading HBM memory chips, but sources of TheElec report that the South Korean memory giant will deliver new HBM to Broadcom in 2H 2025, with the US chipmaker to use the AI memory chips in new AI chips for Google, Meta, and ByteDance.

The deal will see SK hynix push its HBM production capacity to 160,000 - 170,000 wafers in 2025, up from the previous target of 140,000 - 150,000. SK hynix is providing most of the HBM memory chips it makes to NVIDIA for use in H200, GB200, B200, GB200, and other AI chips and AI servers, but now Broadcom wants some of that delicious HBM memory pie for its CSP clients.

NVIDIA is expected to unveil its next-generation GB300 AI server platform at GTC 2025 which takes place between March 17-20, with some huge surprises from the new AI GPUs.

In new leaks from UDN, we're learning that NVIDIA is preparing its new B300 AI GPUs will scream along with up to 1400W of power, with 1.5x the performance in FP4 performance on a single AI GPU over the B200 AI GPU. We can expect the HBM capacity of each GPU to increase from the max of 192GB HBM3E on the B200, to a much bigger 288GB of HBM3E on B300.

NVIDIA's current GB200 AI uses 192GB of HBM3E in an 8-Hi stack configuration, but the next-gen GB300 will use 288GB of HBM3E in a larger 12-Hi stack, which SK hynix has been working on for a while now.

HP's next-gen Omen Max 16 gaming laptop has been leaked, revealing it'll be powered by the new Intel Core Ultra 9 275HX processor and next-gen NVIDIA GeForce RTX 5080 Laptop GPU.

In the new leaks, we're learning the new HP Omen Max 16 gaming laptop will feature the higher-end Arrow Lake-based Intel Core Ultra 9 275HX processor, 32GB of DDR5-5600 memory, a 1TB Gen4 SSD, and NVIDIA's new GeForce RTX 5080 Laptop GPU which will have 16GB of GDDR7 memory and offer stellar gaming performance for 2025.

Intel's new Core Ultra 9 275HX processor features 24 cores (8P+16E) with clock speeds of up to 5.4GHz, with the NVIDIA GeForce RTX 5080 Laptop GPU (should be the GN22-X9 we've been hearing about in recent leaks). NVIDIA's new GeForce RTX 5080 Laptop GPU is expected to roll out with 8192 CUDA cores and 16GB of ultra-fast GDDR7 memory, a maximum TGP of 175W, and more.

NVIDIA's next-generation GeForce RTX 5090 is just weeks away and while we've had plenty of leaks, now it's time for a gander at the PCB of one of the custom RTX 5090 graphics cards.

In a new post on Chiphell, we're being introduced to a beautifully large PCB with over 40 capacitors and 16 GDDR7 memory modules. The PCB we're looking at is for a custom AIB model of the GeForce RTX 5090, with no visible NVIDIA branding on the PCB itself, so this isn't a Founders Edition or NVIDIA reference model, but a custom AIB card.

We've seen leaks of the GDDR7 memory layout and this new PCB leak lines up with that, with 5 modules on each side (4 on top, 2 below the GPU) with 16 modules in total which would allow for some delicious configurations: 32GB, 48GB, or even 64GB depending on whether 2GB, 3GB, or 4GB GDDR7 memory modules are used.