Relativity Networks Raises $22M, Secures $40M Hyperscaler Order
📅 - Relativity Networks has raised $22 million and secured a $40 million follow-on order from an unnamed hyperscaler after testing its hollow-core fiber between two data centers. The company says its ChronoCore technology can cut propagation delay enough to let AI infrastructure spread farther from constrained power hubs, turning network latency into a data center siting variable today.
That is the more interesting part of the announcement. The financing matters, but AI infrastructure operators are increasingly running into a less negotiable constraint than GPU supply: electricity. When adequate power cannot be found beside an existing hyperscale campus, compute has to move. Once it moves far [...][... Check source for end of article ...]
📅 - NTT DOCOMO Expands Quantum Computing in Network Operations - NTT DOCOMO has put a second D-Wave-powered quantum optimization application into production, using hybrid annealing technology to tune mobile network tracking-area lists. The Japanese operator says the system cut peak location-registration signaling by 65.3% and paging traffic by 7%, with a representative optimization covering 333 base stations completed in roughly five minutes during operational network planning workflows.
The interesting part is not that another telecom company has experimented with quantum computing. DOCOMO says the results are already feeding into network planning and daily operations. That moves the discussion away from laboratory benchmarks and toward whether [...]
📅 - Cerebras Launches CS-4 Rack System for Speedier AI Inference - Cerebras Systems has launched the CS-4, a rack-scale AI inference system built around three Wafer Scale Engine 3 Turbo processors and a new modular platform called Nexus. The company claims up to 30 times faster inference than GPU-based systems, twice the performance of CS-3, and deployment times measured in hours rather than days for large AI workloads.
The system is aimed squarely at one of the more stubborn problems in production AI infrastructure: inference speed once models become large enough that moving data between processors starts to matter almost as much as raw compute. Cerebras says CS-4 delivers 750 petaflops of AI compute, 129.6 petabytes per second of memory bandwidth, and [...]
📅 - Cursor Launches Origin to Challenge GitHub in Code Hosting - Cursor has launched Origin, a Git-based code hosting platform built directly into its AI development environment, extending the company beyond coding assistance into repository management and review. The move comes as GitHub faces reliability scrutiny, rising AI-generated code volumes, and enterprise questions over whether developer infrastructure should remain concentrated in one dominant platform for software teams today.
Origin arrives at an awkward moment for GitHub. A recent outage disrupted pull requests, APIs, enterprise authentication, and Copilot for nearly six hours, with error rates approaching 20% across several services and roughly 50% for archive and raw-file downloads. A [...]
