TokenPost topic
AI infrastructure
617 TokenPost stories are tagged AI infrastructure. The newest appear first.
Latest AI infrastructure stories
- MARA Sold 13,597 Bitcoin Worth $1.1 Billion Over Six Weeks
The sales covered about 40% of MARA’s quarter-end Bitcoin holdings. The company used $750 million to repay a Bitcoin-backed loan and release 17,477 BTC from collateral.
· Business
- CoreWeave Plans Customer-Linked Loans as Debt Reaches $46.9 Billion
The AI cloud provider posted $3.66 billion in third-quarter revenue, up 168% from a year earlier, and a $709 million net loss.
· Business
- Nscale Discloses $105.8 Billion in Active and Contracted TCV
The Nov. 12 snapshot covered about 461,000 GPUs active or under contract. Nscale had not obtained binding financing commitments for infrastructure tied to its Anthropic agreements.
· Business
- South Korea Expects Nvidia GPU Deliveries Under Contracts to Continue
About 13,100 Nvidia B200 GPUs have been deployed in the country, and plans to introduce Vera Rubin GPUs are expected to proceed as planned.
· Technology
- Fireworks CEO Says Batch Files Remain Until Customers Delete Them
An account-wide zero data retention setting can reject storage-dependent batch jobs and uploads, but it is off by default and does not delete existing files.
· Technology
- NVIDIA Runs 753-Billion-Parameter GLM-5.3 Across Two DGX Stations
The vLLM setup reached 607.8 tokens per second with 16 requests running at once on the SPEED-Bench Coding dataset.
· Technology
- Digi Power X Targets $250 Million-$300 Million Annualized Revenue Run Rate at Q3 2027 Exit
About $143 million of the goal is tied to current operations and the Columbiana tenant agreement; the rest depends on additional AI colocation and GPU capacity.
· Business
- Roth Starts Analog Devices Coverage With Buy Rating on AI Exposure
Data centers accounted for about 80% of Communications in F3Q26A, when the business grew 84% year over year.
· Investing
- KIDZ AI Posts $654,057 Q3 Loss as Revenue Rises 40% Sequentially
The company has begun initial hardware procurement and deployment planning for its GPU computing infrastructure strategy.
· Business
- CoreWeave Seeks $4 Billion Loan Backed by Equipment and Meta Contracts
The proposed collateral includes CoreWeave's computing equipment and contracts with Meta. The effort follows an $8.5 billion financing round in March.
· Business
- Nebius Sets Q2 2027 Delivery for Remaining Microsoft Capacity
The company expects to deliver more than half of the contracted capacity by year-end and sees 2026 revenue near the low end of its $3.0 billion to $3.4 billion range.
· Business
- Cisco Shares Fall About 2% as Hardware Margins Draw Focus
Cisco beat quarterly estimates and raised its fiscal-year AI infrastructure revenue target to $9 billion.
· Business
- Cerebras Adds Microsoft as Its Third Hyperscaler Cloud Customer
The agreement was announced during Cerebras’ third-quarter earnings call. Core revenue was about $230 million, up 20% year over year.
· Business
- SK hynix Proposes Shared Memory Racks for AI Inference Systems
An internal test using remote CXL DRAM cut average time to first token to 15.45 seconds from 80.76 seconds, an approximately 81% reduction. SK hynix said validation with Intel and partners will start in the system-des…
· Technology
- Jukan Says Future Nvidia CEO Could Pressure Neocloud Margins
Jukan said a pullback in Nvidia’s neocloud support could pressure GPU spot prices and provider margins, pushing some providers to consider AMD or Google TPU chips.
· Business
- Microsoft’s West Texas Data Center Project Targets 2028 Start
Microsoft plans to invest $117 billion in 2026 to build computing capacity for AI and cloud services.
· Business
- Cerebras Falls Behind on Preparing First OpenAI Capacity Tranche as Data Center Deals Lag
The first 250-megawatt tranche is due by the end of 2026 under an agreement covering 750 MW through 2028.
· Business
- Cisco Raises AI Revenue Goal to $9 Billion After Quarterly Earnings Beat
Revenue rose 24% year over year, while Cisco's fiscal second-quarter revenue and adjusted earnings forecasts also exceeded analyst averages.
· Business
- HPE Sets Early 2027 Availability for Cray E2000 Storage Configurations
The Rack Management Controller is planned for March 2027, followed by AMD and NVIDIA rack-scale support in June.
· Technology
- Forum Markets Cuts 2026 Outlook to $8M-$9M, Slows Initial AI Compute Deployment
The company cited rising compute prices, equipment and financing costs as it slowed deployment of initial AI capacity. Its previous 2026 revenue range was $18 million to $22 million.
· Business
- Baidu Unveils Tianchi 1024-card Supernode for AI Workloads
The system uses optical connections to speed data transfers among computing resources.
· Technology
- Astera Labs’ Three Largest Direct Customers Accounted for 77% of Q3 Revenue
The disclosure came alongside record fiscal third-quarter revenue of $584.9 million and a year-over-year decline in GAAP gross margin to 72.3%.
· Business
- Boost Run Reports $39.9 Million in Q3 Revenue, Up 496%
Revenue rose from $6.7 million in the year-earlier quarter. Boost Run also announced more than $600 million in year-to-date financing across multiple projects.
· Business
- CoreWeave Shares Fall 16.8% Intraday to $50.75 on Nov. 12
The intraday quote puts the stock in focus for investors tracking AI and crypto infrastructure companies.
· Investing
- Sam Altman Sees Near-Term AI Demand Swings, Long-Term Token Growth
He cited monthly token use by the top user rising from about 100,000 in early 2020 to 100 billion at the start of 2026.
· Technology
- Nebius Beats Q3 Revenue and Adjusted EBITDA Estimates as Shares Rise Premarket
Revenue climbed 68% from the second quarter, and the company’s earnings call is scheduled for 8 a.m. ET (13:00 UTC).
· Business
- NVIDIA Blackwell Chip Rental Rates Rise 48% in Two Months
Hourly rates reached $4.08 as demand for AI agents grew, while Anthropic and CoreWeave made separate changes to usage limits, prices and contract terms.
· Technology
- CoreWeave Posts $3.66 Billion in Third-Quarter Revenue, Plans Loan of at Least $4 Billion
CoreWeave expects $9 billion in customer prepayments and deposits over the coming months.
· Business
- Meta Infrastructure Chief to Take Advisory Role at Meta Compute
Santosh Janardhan will stay during a leadership transition and is expected to leave Meta in 2027. Dave Brown is slated to succeed him.
· People
- UBS Sees Global Growth Rising to 3.5% in 2027 From 3.2%
Recovering oil and gas output, mainly in Saudi Arabia and the United Arab Emirates, accounts for almost all of the projected increase.
· Investing
- NVIDIA Links GPU Deployment Recipes With AI Workload Inventory
AICR packages validated GPU configurations, while Google-developed k8s-aibom inventories recognized models, runtimes and dependencies in Kubernetes workloads.
· Technology
- Torus Completes $500 Million Series C and Launches Foundry AI System
Foundry’s estimated time from signing to operation is about 12 to 16 weeks. Hydra Host is among its first customers.
· Business
- Meta Infrastructure Chief Santosh Janardhan to Leave in 2027
Meta vice president Dave Brown is slated to succeed Janardhan as the company expands computing capacity for artificial intelligence.
· People
- CoreWeave Plans New Asset-Level Borrowing Over Next Two Weeks
The plan adds to the AI cloud provider’s use of collateralized financing as it pursues data-center expansion plans.
· Business
- CoreWeave CEO Expects No Election Impact on Data Centers Serving Its Backlog
Intrator said some ongoing projects were already permitted during CoreWeave’s Nov. 11 earnings call.
· Business
- CoreWeave Shares Jump 15% as Q3 Revenue Hits $3.66 Billion
The company raised its 2026 revenue forecast to $13 billion to $13.3 billion as quarterly interest expense climbed to $846 million.
· Business
- Google Releases XLA Serving Engine to Deploy Models Without Python
Developers can train models with JAX or PyTorch, then export standalone artifacts for CPUs, NVIDIA GPUs and Google Cloud TPUs.
· Technology
- Tendrils Raises $30 Million to Build a Parallel Processor
Its design represents programs as connected nodes, while Vine is an open-source language intended to support software for the architecture.
· Technology
- AMD Helion Lineup Includes More Than 10 Certified Rack Models
The configurations include liquid-cooled and air-cooled CPU racks, plus a mixed CPU-GPU system designed for enterprise inference.
· Technology
- Anthropic Says Claude Mythos Ported Claude to AMD GPUs in About Three Days
A separate partnership calls for Anthropic to use up to two gigawatts of AMD Instinct MI450-series GPUs, with deployment planned for the first half of 2027.
· Technology