GPUs

Evidence (10)

• 2Hacker News·7/29/2026open_source
  Darktable Open-Source Photo Workflow App

  Darktable is an open-source photography workflow application and raw developer for managing digital negatives and editing images non-destructively. It includes a virtual lighttable and darkroom, color management features, and GPU-accelerated processing for photographers.

• 4Hacker News·7/28/2026open_source
  XY: Fast GPU-Accelerated Interactive Plotting Library

  XY is a newly shared open-source plotting library focused on fast, composable, interactive chart rendering. It uses GPU acceleration to improve performance for data visualization workflows and targets developers who need responsive plotting in the browser or application UI.

• 9Hacker News·7/21/2026funding
  Hidden AI-related debts at US tech giants surge to $1.65T

  A Nikkei study says off-balance-sheet liabilities at five major U.S. tech companies have risen sharply as AI spending expands, reaching an estimated $1.65 trillion. The report says data center leases and GPU supply contracts are making the companies' true obligations harder for investors to see, with Meta and Oracle highlighted as examples.

• 7The New Stack·7/15/2026product_launch
  Meta moves to monetize excess GPU capacity

  Meta is reportedly building a cloud business under its Meta Compute initiative to sell excess AI infrastructure, with possible offerings ranging from hosted AI model access to raw compute rental. The article frames this as part of a broader shift where overprovisioned GPU capacity is being turned into cloud supply, increasing fragmentation and operational complexity for buyers.

• 7Hacker News·7/14/2026product_launch
  Spectral Compute Pushes a CUDA-Free Path for Non-NVIDIA GPUs

  The article covers Spectral Compute’s effort to make CUDA-style GPU programming work on non-NVIDIA hardware, potentially reducing dependence on NVIDIA’s software stack. The Hacker News discussion reflects interest in whether a credible alternative can emerge for developers and HPC users who want broader hardware compatibility.

• 9The New Stack·7/10/2026breakthrough
  Meta readies first in-house AI chip Iris

  Meta is set to begin production of its proprietary AI chip, Iris, in September after bug testing, with the chip expected to offload some inference workloads from third-party GPUs. The move is part of Meta’s broader push to build more of its AI infrastructure in-house, cut costs, and reduce dependence on external hardware suppliers.

• 5InfoQ·7/10/2026research
  Chaos Engineering for GPU Clusters

  Bryan Oliver presents practical approaches to chaos engineering for large-scale GPU clusters used in AI infrastructure. The talk focuses on handling complex topologies, RDMA networking, NUMA misalignment, and fault injection to improve hardware efficiency and observability.

• 7The New Stack·7/2/2026security
  Cargo theft emerges as an AI infrastructure risk

  Authorities recovered two stolen trailers carrying about $1.3 million in data center equipment and copper wiring outside Chicago, underscoring a physical supply-chain risk for AI infrastructure. The incident highlights how theft of servers, networking gear, power equipment, and copper can delay data center construction just as AI deployments accelerate.

• 7The New Stack·7/1/2026product_launch
  Clockwork launches a guarantee to avoid AI training restarts

  Clockwork introduced its YOCO Guarantee, promising that 90% of failures on supported training runs will be recovered without losing progress, rolling back checkpoints, or recomputing work. The offer is built on TorchPass, a fault-tolerance product that can live-migrate GPU training state to a healthy spare or lower-priority node, with recovery ranging from seconds to a few minutes depending on the mode used.

• 6The New Stack·6/23/2026research
  Kubernetes teams trust automation for delivery but hesitate on CPU and memory tuning

  A survey of 321 enterprise Kubernetes practitioners found strong trust in automated deployment and rollback, but much lower willingness to let systems auto-apply CPU and memory resource changes. The gap matters more as AI inference workloads on Kubernetes raise the cost of overprovisioning and make manual rightsizing harder to sustain at scale.