TokenPost topic
AI infrastructure
618 TokenPost stories are tagged AI infrastructure. The newest appear first.
Latest AI infrastructure stories
- NVIDIA Introduces DOCA GPUNetIO to Unify GPU Networking
The software development kit lets CUDA kernels handle networking and data movement with less CPU involvement across NVIDIA’s software ecosystem.
· Technology
- AI Memory Constraints Drive Tiered Storage and CXL Designs
Vera Rubin systems pair HBM4 with LPDDR5X, storage and high-speed links as longer AI context windows increase memory requirements.
· Technology
- NVIDIA Introduces Multi-GPU Solver for 100M-Variable Models
The cuOpt solver uses NVLink-connected GPUs to handle models with up to 2.1 billion nonzero matrix entries while cutting per-GPU memory use by up to 6x.
· Technology
- NVIDIA Expands DSX Air With Digital Twins and AI Agents
The platform lets operators test AI-factory infrastructure, software integrations and operational changes before physical hardware is deployed.
· Technology
- NVIDIA Tests Robots to Assemble Tester Trays for GB300 AI Systems
The experimental setup handles busbar fastening and four-connector insertion, but the systems have not yet reached factory deployment.
· Technology
- TSMC September Revenue Rises 54.6% as Record Streak Continues
Revenue reached $16.03 billion, extending TSMC’s record monthly sales streak despite a 0.6% decline from August.
· Business
- AI Power Bottlenecks Could Shift Risk Across the Chip Supply Chain
U.S. electricity demand is rising as data-center projects face constraints involving power, grid connections, land, permits and completed facilities.
· Technology
- Spiral Develops Bitcoin Lightning Payments for Mesh LLM AI Compute
The planned system would let AI agents pay machines providing inference, but no live payments or paid production use have been confirmed.
· Technology
- Arista Unveils Ethernet Architecture for AI Racks With 1,024 XPUs
The Etherlink SU-144 architecture connects up to 144 accelerators in a single-hop domain and up to 1,024 XPUs through cross-rack topologies.
· Technology
- Alibaba Chairman Tsai Urges Europe to Expand Sovereign AI Infrastructure
Tsai called for more European data-center capacity and open or open-weight models to reduce reliance on technology controlled by the United States and China.
· Technology
- Forum Markets Forms U.S. AI Venture With 11-Megawatt Plan
The majority-owned venture with Edge Node Inc. will deploy GPU capacity at powered data centers and mobile-tower sites, with 2027 revenue guidance of $125 million to $175 million.
· Business
- U.S. Data Centers Could Face 32 GW Power Gap Through 2028
The projected shortfall could shift bottleneck risk from chips to electricity and delay orders for memory, optical and power-management products.
· Technology
- AMD Moves Helios AI Racks Into Production Ahead of 2026 Shipments
Each rack combines 72 MI455X accelerators and 18 EPYC CPUs, with deployments planned by Microsoft, OpenAI and Anthropic.
· Technology
- TSMC Expands U.S. Investment to $265 Billion as AI Supply Chains Shift
The company plans four more advanced fabs in the United States, while Taiwanese technology companies expand AI-server supply chains across Southeast Asia.
· Business
- Norway’s ASP Data Center Backlog Nears 2 Billion Kroner
A 6-megawatt customer agreement brought contracted revenue backlog close to 2 billion Norwegian kroner. ASP also outlined projects in Norway and Finland, including a planned 30-megawatt first phase in Finland.
· Business
- Wingspire and Stonebriar Package Equipment Contracts Into Securities
Wingspire’s notes total $407.07 million, while Stonebriar’s pool has an initial discounted contract balance of about $957.87 million. Neither public summary specifies the share tied to AI computing.
· Business
- HIVE Puts Two GPU Cloud Deals at $570 Million and Plans GTA Site
The agreements cover 4,320 GPUs. HIVE’s proposed Greater Toronto Area facility is designed for more than 100,000 GPUs at full build-out.
· Business
- Firmus Plans $5.5 Billion IPO at $30.6 Billion Valuation
The Australian AI infrastructure company expects the offering to be the country's second-largest IPO, with about half the shares reserved for selected existing investors.
· Business
- Nebius Integrates Inferize Technology Into Token Factory
The acquisition adds engineering work focused on reducing delays when AI models start and scale. The deal’s financial terms were not disclosed.
· Technology
- Akamai Trades at Lower Earnings Multiple as AI Traffic Grows
Fastly shares have risen 159% this year, while Akamai has gained 22%. Jim Cramer favored Akamai among the three companies because of its lower valuation.
· Business
- ChronoScale Reports $71.6 Million in Fiscal 2026 Revenue
Together AI accounted for all of the revenue. ChronoScale also reported a $49.3 million loss from continuing operations before income tax.
· Business
- Lagarde Warns Foreign AI Dependence Could Disrupt European Industries
The ECB president called for more computing capacity and AI models running on European infrastructure as the United States leads in computing resources.
· Technology
- HPE Secures $1.2 Billion Vultr Order for AMD Helios AI Systems
The U.S. deployment includes HPE networking hardware and software. The order comes as HPE sets separate targets for networking revenue growth and operating margins.
· Business
- Micron Earnings Put HBM Demand and the Memory Cycle Under Scrutiny
The company’s fiscal fourth-quarter guidance trails consensus estimates as investors weigh artificial-intelligence demand, conventional DRAM pricing and fiscal 2027 spending.
· Business
- South Korea’s September Exports Hit Record $120.944 Billion
Exports rose 83.5% year over year, while semiconductor shipments surged 262.8% to $60.3 billion amid strong memory-product demand.
· Business
- Micron Sees Tighter Memory Supply Through 2028 as AI Demand Drives Spending
The chipmaker expects $61.5 billion in fiscal first-quarter revenue and plans about $25 billion in capital spending during fiscal 2027’s first half.
· Business
- Micron CFO Projects $61.5 Billion Revenue for Fiscal 2027 First Quarter
Mark Murphy forecast earnings of $38.15 per share and said the quarter would mark the low point for fiscal 2027 gross margin.
· Business
- GMI Cloud Completes $668 Million Financing With NVIDIA Participation
The financing includes $223 million in equity and $445 million in credit, with ARCHIV leading the equity round.
· Business
- Arthur Hayes Points to Potential AI Credit Stress in 2027–2028
The Maelstrom chief investment officer says a slowdown in AI infrastructure spending could prompt government liquidity creation and drive Bitcoin to $1 million or more.
· People
- Alphabet Begins Recognizing Revenue From TPU Systems as Google Cloud Backlog Reaches $513.9 Billion
Google Cloud began recognizing TPU system sales in the quarter ended June 30, while TSMC described rising demand for AI capacity and advanced packaging.
· Technology
- CoreWeave Launches Forge to Expand Beyond GPU Infrastructure
The platform connects model serving, monitoring, data curation, training and evaluation across models, frameworks and cloud environments.
· Technology
- AI Infrastructure Spending Pushes Investors Toward Cash Flow
Microsoft, Oracle, Meta and Tesla are committing hundreds of billions of dollars to data centers, computing capacity and other AI-related assets.
· Investing
- CoreWeave Puts NVIDIA Vera Rubin NVL72 Into Production for Cloud Customers
Cognition’s early tests found up to 4.8 times the total token throughput of NVIDIA’s GB200 NVL72 in software-engineering AI inference tasks.
· Technology
- Cerebras CEO Andrew Feldman To Discuss AI Scaling Limits Oct. 13-15
Feldman will address rising demands for computing power, electricity and data-center infrastructure at TechCrunch Disrupt 2026 in San Francisco.
· Technology
- Restate Raises $20 Million to Expand Durable AI Infrastructure
The Series A will fund hiring and expansion in the San Francisco Bay Area as the Berlin-based startup takes on Temporal.
· Business
- Nebius Signs AI Capacity Deal With Meta Worth Up to $27 Billion
The five-year agreement includes $12 billion in dedicated capacity starting in early 2027 and up to $15 billion in additional purchases.
· Technology
- Bitdeer Secures More Than $1.7 Billion in Johor AI Contracts
The five-year agreements cover more than 70% of the facility’s 21.7-megawatt capacity. Bitdeer expects total revenue backlog to reach about $2.9 billion.
· Technology
- DeepSeek and Huawei Expand Ascend Support Across AI Software Stack
New open-source tools cover operator development, matrix computation, cross-chip communication, Attention and data selection as the companies reduce reliance on Nvidia’s CUDA ecosystem.
· Technology
- Hut 8 Secures $1.07 Billion Revolving Credit Facility
The four-year facility was undrawn at closing, with liquidity requirements of 40% before the Stabilization Date and 25% afterward.
· Business
- Nvidia Targets More Than $500 Billion in Third-Party AI Capital
Nvidia signed memorandums of understanding with six financial firms, but the arrangements may not lead to definitive agreements.
· Technology