US Export Curbs Hit Server Makers
New US limits on advanced AI chip exports cause shares of HPE, Dell, and Super Micro to slide.
ddNews Β· news about Hewlett Packard Enterprise (HPE) β products, AI,
New US limits on advanced AI chip exports cause shares of HPE, Dell, and Super Micro to slide.
HPE and Dell saw stock dips as Washington tightened rules on AI chip shipments to China.
The FAA immediately halted new installations after a network upgrade triggered massive flight disruptions.
The Federal Aviation Administration halts deployment following a nationwide ground stop investigation.

The company has created a 1,040-qubit quantum system.

A $1.7bn deal allowing Millicom to enter the Peruvian market.

Strong investor demand is pushing the share sale price toward the top of its estimated range.
The sparse-native computing leader adds veteran semiconductor and HPC leadership to its board.

The Chinese commerce ministry opposes new US limits on remote access to advanced AI chips.

CBRE predicts Milan will overtake Dublin by 2028 with 255MW of new capacity.

Nvidia forms a board committee to oversee export controls and national security amid China scrutiny.

New capabilities in Qlik Answers improve AI-fueled insight generation through custom skills.

The pair are expanding infrastructure sharing to over 1,000 sites across Japan.
HPE's VP of Private Cloud AI shares insights on the company's strategic direction in an interview series.

Vulnerability disclosure timelines are being shortened due to AI-accelerated exploitation.
The Federal Aviation Administration has halted the rollout of Juniper routers following a nationwide ground stop.

The software update makes language models on Modalix chips up to 3.5x faster.

Google will buy 50MW of "superhot" geothermal power from Quaise Energy starting 2030.
quality(80)/p7i.vogel.de/wcms/d2/c3/d2c3f63440ed27171a979cc15d3a7db0/0134553956v1.jpeg) β New restrictions aim to prevent Chinese entities from accessing modern Nvidia chips via loopholes.

New integration standards aim to improve reliability and request handling across server instances.
The agency took action after a nationwide ground stop, though the root cause remains unknown.

The Alphabet drug discovery company secured a Series C round co-led by Sequoia and Greenoaks.

The new DRIL lab allows APAC enterprises to validate AI-ready infrastructure.

A joint venture will develop two major projects, including Egypt's first sovereign AI facility.

Lumen highlights legacy network issues while HPE rejects blame regarding Juniper-related networking failures.
The new system debuts among the world's 15 fastest as part of a $1bn partnership.

AI energy shortages are forcing a shift toward behind-the-meter generation and power-first site selection.

The deployment utilizes Nvidia's optical networking architecture at a massive production scale.

A Harvard study finds the 403 largest US data centers use 1.8% of national electricity, mostly from fossil fuels.

Early MLPerf results indicate Nvidia Rubin is up to twice as fast as Blackwell for training.

The company aims to fund AI data centers across Africa.

Beijing is evaluating countermeasures ahead of the Trump summit regarding AI chip access.
quality(80)/p7i.vogel.de/wcms/03/12/03123962b6599bee9b6f20395fc931f1/0134524477v1.jpeg) β Analysis suggests six Chinese AI models were initialized using GPUs that bypassed US export licenses.

A troubling milestone where AI agents move from assisting hackers to executing attacks independently.

Dr. Fei-Fei Li joins AMD as EVP and Chief Scientist to expand AI capabilities.
The new system is part of a $1 billion public-private partnership for the US Genesis Mission.
U.S. senators are urging the FAA to suspend Juniper deployments following a nationwide airline ground stop.

A six-year deal to deliver over 450MW of AMD-powered AI infrastructure.

Google reports the first instance of attackers using AI to turn patches into exploits without human intervention.

The state alleges rogue AI agents hacked Hugging Face, endangering the public.
The system is now open to scientific users at Oak Ridge National Laboratory.
A joint effort between AMD, ORNL, and HPE to provide high-performance computing for critical research.
The new cluster developed for Oak Ridge National Laboratory debuts as a top-tier AI and HPC resource.

Average cores per system hit 445,441, driven by hybrid GPU-based accelerators.

The funding will accelerate tools for companies to build and run enterprise AI agents.

A specialized version of the Rubin GPU featuring over 50,000 CUDA cores optimized for scientific computing and data processing.
The company's share price saw a significant decline on Monday evening.

Data shows a dive in new supercomputer installations during the second half of 2026.
Investors sold off HPE stock on Monday afternoon.

Regulations now pressure overseas data center operators to verify who uses their computing capacity.

The Department of Commerce added five countries, including Malaysia and Singapore, to its export restriction list.

A new AI lab founded by Stanford professors focusing on self-improving AI for the enterprise.

The new accelerator features over 50,000 CUDA cores designed for agentic AI in science and engineering.

A six-year agreement with TensorWave will provide massive AI compute across multiple sites.
Stocks dropped following US government moves to shut down specific AI chip export loopholes.
Stocks dropped following a US Commerce Department move to close a loophole regarding AI chip exports.
Experts argue that strong AI and enterprise infrastructure demand could offset weakness in the PC market.

HPE states the Arkansas router involved in the billion-dollar aviation outage worked as configured.

Long-term models project a potential average price of $91.68 by 2030.

New open datasets and recipes help researchers adapt Nemotron 3.5 for life sciences and chemistry.
A federal probe into smuggling AI servers to China is impacting sector valuations.

The new model features 2.4 trillion parameters and a one-million-token context window.
Sandeep Pendharkar outlines strategies for ransomware recovery and governing AI agents.
Market options indicate significant price movement expected following recent financial figures.
Analysts are closely watching the upcoming fiscal Q4 release to see if AI demand translates to revenue.
The bank reaffirms its positive outlook on Hewlett Packard Enterprise.

The use of Zutacore's two-phase liquid cooling is increasing HPC capacity while reducing energy.

A multi-site federated project across the Nordic and Baltic regions to ensure sovereign cloud and AI capacity.
Despite the slide, Morgan Stanley and Citi suggest strong AI demand provides earnings support.

Underground brokers are selling hijacked enterprise AI account access, shifting costs to the companies.

Analysts warn CISOs to implement strict rollback and recovery plans as AI agent guardrails fail.
The company introduced what it claims is the world's fastest hybrid storage technology for supercomputing racks.

New models have reduced the cost of finding critical vulnerabilities to as little as $4.

"DirectROUTES" allows smaller networks to access the Telekom network via existing DE-CIX connections.
The technology delivers up to 3x higher token throughput for AI compute.
New 102.4T switches and 800G Thor Ultra fabrics are presented for AI Ethernet stacks.
New 800 VDC portfolio aims to solve thermal design challenges for multi-megawatt AI clusters.

Four new AMD EPYC-based servers are now available, specifically targeting agentic AI and high-performance analytics.