Computex 2026 also sees firm launch 102.4 Tb/s Ethernet switch

Air-cooled GPU for inference workloads

The $750 million deal sees Credo expand its portfolio as it eyes to offer a ‘vertically integrated connectivity stack’

South Korean startup’s third-generation accelerators will have a multi-die system architecture

Tensormesh now worth $24.5 million and counting

Huawei’s τ Scaling Law argues the future of chips lies in time, not transistor size

AMD, Everpure, Keysight, Viavi and more join neo-network venture

Announcements span local AI agents, rack-scale infrastructure, data orchestration, and partner ecosystem integrations

SpaceX gets its hands on Vera CPU release after OpenAI defeat

How about security and Splunk? Well ... it's complicated

FortiAIGate refresh broadens pair's partnership

The pair injected network efficiency tech into 'large-scale commercial trials'

DoCoMo performance sees Japanese giant push back targets and go all in on IOWN APN

Fresh off deals with Meta, Google, and Anthropic, chip giant eyes funding to fuel custom silicon production

Zerogrid routes inference workloads according to latency, locality, and capacity constraints

Optics supplier posts $1.81 billion in Q3 revenue and is on track to quadruple its indium phosphide production capacity amid supply rumblings

Chip giant extends accelerator peripheral lineup as it looks to meet bandwidth-booming workload demands

AI giant offers rare look under the hood with MRC

'We've done more in six months than, for example, Cisco Meraki has done in over 13 years to bring those together,' HPE's Jeff Aaron claims

Scorpio X-Series 320 looks to minimize latency in large AI clusters, while P-Series expands to broaden PCIe 6 switch options

Liquid cooling hire comes as reports suggest ConnectX-9 power consumption is already causing headaches

There's also a new 'ruggedized kit'

Begun, the AI pricing wars have

Vendor giant deepens relations with edge chip designers with expansive APAC ambitions