Systems news | The Register

REG AD

ON-PREM > SYSTEMS

Latest news and analysis of datacenter server systems

Datacenters face direct hit from China rare earth curbs, as clock runs out on escalated licensing chokeoff

Paused export controls on materials used in key IT components due for review in November

Nvidia and Cerebras are selling performance their customers will (probably) never see

Touting batch 1 token generation is a bit like boasting about the top speed of your car

A closer look at Meta's MTIA 400 accelerator package

Meta's new MTIA 400 chip has a split personality: Training AI and serving ads

Faster than Blackwell, but still no replacement for AMD or Nvidia ... yet

MNT follows up its open hardware portables with a desktop case

Upgrade the portable, give its old mainboard another job

REG AD

Intel's 256-core Xeon 7 CPUs are a Diamond in the rough

Delayed, flawed, yet still beautiful

OpenAI's upcoming Jalapeño chip looks like it'll be an inference beast

128 chips, 1.7 exaFLOPS, and 27 TB of HBM give Altman and crew a leg up over Blackwell, and maybe even Rubin

SiFive pushes development server to take RISC-V into the datacenter

Rack-mount system intended for software porting, workload optimization, to bring open standard to the bitbarn

IBM announces chip that natively executes Arm and Z instructions concurrently

Cunning plan to expand the amount of software that runs on mainframes is also an admission ISVs are focusing their energies elsewhere

YouTuber's backyard fab ambitions expand from RAM to homebrew LEDs

Dr. Semiconductor's latest project lights the path toward what he tells us he hopes becomes a cottage chip industry

What Nvidia's first Groq 3 LPU benchmarks tell us about its $20B gamble

Gemma 4 31B performance tests offer a best-case scenario for next-gen dataflow accelerators

Macro Shot of a Silicon Wafer with Computer Chips during Manufacturing Process at Fab or Foundry. Semiconductor Wafer Texture.

Chipmakers laughing all the way to the vault as memory prices go stratospheric

Gartner reckons $1.6 trillion year is just the start, and don't expect a bust anytime soon

Cloverleaf deal is latest example of Nvidia using its war chest to patch cracks in the AI bubble

Nv's new chips need new datacenters, but you can't have bit barns without power

REG AD

AMD grabs more CPU share while pricier PCs punish desktop demand

Mercury Research blames costly memory and scarce GPUs for 20% processor shipment slide

IBM says super-chill boxes that connect through 'cryogenic tunnels' will get quantum computers scaling

Absolute zero inside is much cooler than Intel inside

AMD inches closer to its goal of making AI suck less ... energy

House of Zen claims latest systems already 4x more efficient than two years ago

Baidu says Chinese buyers want local AI chips due to ‘supply chain’ issues

It seems like Nvidia’s not getting back into the Middle Kingdom anytime soon

Cerebras CS-4 rack systems juice chips for every last drop of AI performance

Next-gen systems double per-chip performance while cramming 3x as many into a rack

Part didn't fit so techie got out his screwdriver. Then something flew off the motherboard

Whatever it was broke a PC and led to a face-saving lie

Big Cloud is poised to corner the market for enterprise hardware

AI gave hyperscalers first dibs on scarce kit, and business buyers may have little choice but to rent it back

Intel upsizes stock sale to $20B with spending plans still fuzzy

Semiconductor giant says opaquely it will use the proceeds for 'general corporate purposes'

London still dominates Britain's datacenter map, but the regions are powering up

The capital holds two-thirds of capacity, while multimegawatt projects gather beyond the M25

AMD acquires AI chip startup Taalas to boost inference performance by etching models into silicon

Early tech demos show model-specific integrated circuits churning out up to 17,000 tokens a second

Snowflake extortionist admits 165-victim cloud crime spree – and squeezing one target twice

Connor Moucka pleads guilty over sprawling 2024 campaign that looted billions of records

HPE extends validity of quoted hardware prices, possibly for months

Suggests Big Green knows its component prices will remain steady

Elon pledges to give Nvidia a virtual monopoly over the stars

In space, no one can hear you scream when the LLMs hallucinate

AMD's AI eggs are in too few baskets, Wall Street worries

The House of Zen's new Helios racks, Venice Epycs, may dent Nvidia's dominance — if the bubble doesn't pop first

A deep dive into Nvidia's Vera CPU and the Olympus cores that power it

88 custom cores, 176 funky threads, 1.5 TB of laptop RAM, and 1.8 TB/s of NVLink connectivity — this isn't your typical datacenter chip

UK wants datacenters to pay a fee for grid connection requests

Refundable charge intended to discourage time wasters from filing applications that will never be realized

Close-up of green RAM memory modules arranged in a pile

Samsung warns memory crunch will last through 2028 as profit rises 19-fold

Tech giant counts record piles of cash while buyers face years of elevated prices

Cristiano Amon speaks onstage at a Qualcomm AI keynote, holding a device.

Qualcomm won’t be a big datacenter player anytime soon

But AI has come along just in time to cover being dumped by Apple

REG AD

Big Tech demanding deals that smooth out memory prices, says SK Hynix

Don't worry about an AI bust, monster profits and huge margins are here to stay

Intel knows it needs to ‘leapfrog’ ARM and AMD, says CEO Lip-Bu Tan

Chipzilla changes PC biz name to reflect belief edge and robots could be as valuable one day

AMD and Cerebras join forces against Nvidia’s Groq LPUs

The enemy of my enemy is my friend

AMD attacks the rack with Helios systems that rival Nvidia's

Spec for spec, the House of Zen's first rack-scale AI compute platform is bigger and faster than Nvidia's Vera Rubin by nearly every metric, but that's only on paper

Nvidia shows off Vera Rubin platform for tokenmaxxing

If your AI Factory sells tokens, why not optimize token emission?

Intel fortifies Foundry with an actual customer: Fortinet

Firewall maker looks to safeguard its custom ASIC production with homegrown silicon

REG AD

Scalding hot AI accelerators have put datacenters in hot water. Two-phase cooling could chill them out

Startup Accelsius claims it can achieve up to 14° C cooler GPU temps by by converting a liquid-cooled Dell PowerEdge to run on refrigerants

Windows Server Update Services buckle under Microsoft's metadata mountain

Redmond has stabilized new installations, but existing deployments remain stuck in sync purgatory

High hardware costs see Red Hat offer a two-server edge rig, no mini-PC required

Worries about ‘prohibitive cost of powering, maintaining, and deploying a third node’

Chinese memory ban would cut off RAMpocalypse relief

Two US lawmakers push tighter curbs on chipmakers from the Middle Kingdom

TSMC's $265B US fab pledge is the outline of a concept of a plan

Beware of fab makers bearing press releases

Big Blue thinks small, again, with POWER tower

The last proprietary minicomputer, now in ‘deskside’ form if you fancy that

HP laptop showing the Windows 11 desktop and taskbar on screen.

Microsoft flips Windows Backup to on by default unless you're in the EU

Everyone else must opt out manually if they don't fancy settings data shipped off-device

Samsung’s profits jump 19x in a year and you don’t need AI to figure out why

Share price down sharply, apparently amid fears the bubble is getting bigger

IBM teases new rackable mainframes that ‘complete’ the z17 family

UPDATED: Big Blue happy for users to BYO rack and slot Z machines alongside other tech

Madlad builds homebrew GPU using 8,192 RISC-V chips

The next version will have 32,000 MCUs

Ampera CEO Brian Matthews stands in front of the company's full-scale 3D printed nuclear reactor module.

Startup targets datacenters with 3D-printed nuclear reactor module

Fancy a thorium microreactor capable of delivering up to 30 MWe of juice for up to 30 years?

Qualcomm's proposed solution to catch up in AI infra: Bury the compute under the DRAM

With its next-gen AI accelerators, the SoC vendor aims to fly high above the memory wall

Zuck saves Meta bucks by reusing memory from old servers with a custom CXL ASIC

In production on millions of boxes and the payoff is a 25% reduction in machines needed for some inference workloads

Engineer accused of insider trading tied to Microsoft's reboot of Three Mile Island nuclear plant

SEC claims former Constellation employee made $1.4M trading options before the restart deal was announced

REG AD

Man holding a semiconductor wafer manufactured using IBM's new sub-1 nm process

IBM stacks up a sub-nanometer chip future

Big Blue shows off process node it claims can scale down to 1 Angstrom

Hands holding a DDR3L RAM module in front of a blurred background.

Micron locks in historically high memory prices for five years

Big buyers agree to deals that will deliver historically enormous margins and profits

Qualcomm claims it's not too late for Dragonfly to land in datacenters

Oh, Snap(dragon): DC chief says the mobile-chip giant sees bit barns as its next growth market

OpenAI gets chippy with Broadcom

Jalapeño is the latest announcement that attempts to portray OpenAI as more than a race-to-the-bottom model maker

British Home Office seeks techie to herd nearly 1,000 engineers looking after 600 systems

£120k ... but you must take ultimate responsibility for functionality of e-gates, passports and more

Datacenters dip a toe back into waterborne computing despite obvious challenges

Floating or sub-surface bit barns are all the rage, but unlikely to compete with multi-gigawatt sites

Nvidia gets all agentic about supercomputing for scientific research

Tireless AI agents could help scientists do research humans alone can't, says GPU giant

Tensordyne makes a big bet on log math to beat Nvidia

Who needs compute-hungry multiplications when you can just add logarithms

Nutanix's Tech Day London 2026 offers infrastructure insights

SPONSORED POST: Come join this working afternoon for infrastructure teams

Uncle Sam bets $500M that Alphabet spinoff's AI can dig up new semiconductor materials

AI drug discovery is so last year, even though it hasn't accomplished much yet

Intel starts cooking up enhanced 18A-P silicon for would-be foundry customers

Chipzilla claims 9% speed bump without extra power draw but is compatible with designs for 18A

AMD's Mext buy shows how AI could solve the RAM shortage it created

Running low on memory, can't afford more? The House of Zen's latest acquisition puts an AI spin on flash-based memory expansion

There's no such thing as an agentic CPU

AI agents are a general-purpose workload no different from any other

Microsoft faces down sueball, capacity problems in series of challenges

Misleading statements about Copilot and AI? Surely not!

Non-x86 servers now nearly half the market, IDC says

Demand for AI systems plus the shortage of DRAM and NAND are shaping the global market

Shark fin emerging from blue water near the surface.

Qualcomm said to be circling AI chip biz Tenstorrent in $10B RISC-V power play

Potential takeover would represent significant commitment to the open instruction set architecture

SK Hynix to boost memory production 3x ... you can wait another 8 years, right?

We're moving as fast as we can, says SK Group chair

Here's a look at Delos Data's reference design. The front of the system features 36 OSFP cages each capable of 1.6 Tbps of bandwidth.

Delos Data offers AI chip startups a fast track to rack scale

Half the trouble of building an Nvidia NVL or AMD Helios competitor is just getting the networking out of the box

Qualinx QLX3xx chip held in tweezers above a circuit board

Dutch chip startup claims all-European fab flow – with help from a very American friend

Satnav parts designed and manufactured in the EU, but using GlobalFoundries to produce them

Oracle's AI datacenter splurge gives investors the capex jitters

Q4 sales climbed 21%, but Wall Street more interested in $70B buildout bill