Arista's Q3 Results Point to the Network as a Core AI Infrastructure Layer
Arista's Q3 revenue jumped 58.5% and guidance rose again. The numbers suggest networking is becoming a performance decision in AI clusters, not a commodity purc...
Latest stories
Arista's Q3 revenue jumped 58.5% and guidance rose again. The numbers suggest networking is becoming a performance decision in AI clusters, not a commodity purc...
Qualcomm missed on revenue and guided below Wall Street, while Arm and GlobalFoundries beat estimates by thin margins. Investors still sold off or sat flat.
Phaze, a remote desktop startup from Parsec's founders, raised $8.75M in seed funding led by a16z to let people and AI agents share high-end workstations remote...
Memory shortages and long delivery times are pushing Locaweb to plan infrastructure earlier and test alternative hardware configurations with Dell.
Storage is emerging as a bottleneck for enterprise AI, with checkpoint speed, data delivery and unstructured data access now shaping GPU utilization.
Google adds TPU 8i Max for million-token context workloads and TPU 8tx for confidential inference, extending its eighth-generation TPU family.
Staera's $4.6M pre-seed round funds Lightwire, an on-chip optical interconnect it says could replace copper links between AI accelerator chiplets.
AMD's EPYC Embedded 9006 SP7 chips bring up to 256 Zen 6 cores to chip test rigs, firewalls and storage, with PCIe Gen 6 and fast DDR5 memory.
AMD's Venice-generation Epyc 6 lineup pairs 256-core density with 5.3 GHz parts, aimed at running large numbers of AI agents on CPUs.
Enterprise AI buyers need networking, shared inference caching and open-source-friendly stacks, not just accelerators, according to Vultr's CMO.
Fractile raised $720M at a $7B valuation for inference chips that stack DRAM directly on logic to cut data movement. Its performance claims are still unshipped...
Luminal raised a $23M Series A to run AI models across non-Nvidia chips without weeks of manual rewrites, a bet on the software layer behind Nvidia's dominance.
Nebius beat Q3 revenue and profit estimates, but a steeper spending plan and a full-year forecast below consensus sent shares down 17%.
Cerebras reported $230.9 million in core adjusted revenue, up 70%, and a backlog of $26 billion. Shares rose nearly 18% in late trading.
Cisco beat first-quarter estimates and now expects AI infrastructure revenue of $9 billion in fiscal 2027, more than double fiscal 2026. Margins slipped.
Applied Materials reported $10.41 billion in quarterly revenue, up 53%, and guided above forecasts as AI-driven chipmaking equipment demand keeps climbing.
Stack Overflow's survey ranks PostgreSQL first for a fourth year as AI coding agents drive new demand for the database.
Agents drive far more load and token spend than chatbots did. Couchbase argues caching, memory and fewer data hops can cut costs without downgrading models.
Prophecy's DataGPT runs data workflows inside ChatGPT, Claude and Copilot, saving each step as versioned code that IT can review and schedule.
Monte Carlo Data's Bayes monitors deployed AI agents and their data pipelines, diagnoses failures and applies fixes. Early results are company-reported.