by Chris DePuy, 650 Group / October 14, 2026 This morning, Marvell announced the second-generation Structera X 3406 and 3506 CXL 3.2 memory expanders — the industry’s first six-channel CXL memory expanders aimed at AI inference — as the OCP Global Summit gets under way in San Jose. The company… Read More »Marvell Doubles Down on the AI Memory Wall and the Wires Around It
by Chris DePuy, 650 Group / October 9, 2026 This week, Broadcom announced its 2026 OCP Global Summit slate, placing its AI networking lineup — the Tomahawk 6 and Tomahawk Ultra switch families with Jericho 4, Thor NICs at 800G (Thor Ultra) and 400G (Thor 2), and the third-generation TH6-Davisson… Read More »Broadcom’s OCP Slate: Tomahawk 6 Davisson CPO and Ethernet for 100,000-Accelerator Clusters
by Chris DePuy, 650 Group / October 7, 2026 At the OCP Global Summit, which runs October 12–15 in San Jose (OCP Global Summit), we joined Cisco’s data center team for pre-show analyst sessions with Will Eatherton, SVP of Engineering for Data Center & Internet Infrastructure, and Rakesh Chopra, SVP… Read More »Cisco Charts the Pluggable Optics Fork: USFP versus XPO
by Alan Weckel, 650 Group / October 5, 2026 The full interview between 650 Group’s Alan Weckel and Will Eatherton, SVP of Data Center, Internet & Cloud Infrastructure Engineering at Cisco, went up on the 650 Group YouTube channel a few months ago. It was such an interesting conversation that… Read More »AI Backend Network Fabric: Lessons from Cisco Live
Marvell has significantly increased the company’s short-term and long-term guidance during its investor day in NYC. CEO Matt Murphy, followed by the rest of the executive team, were excited to share how Marvell’s TAM would reach $400B by 2030, supported by nearly $3T of CAPEX. The four main pillars of… Read More »Marvell Investor Day – The Path to Multi-Trillion CAPEX and Tokenomics
by Chris DePuy, 650 Group / September 30, 2026 Today, we attended HPE’s Networking Investor Day, where the company put numbers on its post-Juniper networking strategy — and, the same morning, backed the story with its first AMD Helios AI Rack order, a $1.2B deal with Vultr. The through-line of… Read More »HPE Networking Investor Day: Raised Synergies, Segment Growth Targets, and a $1.2B AMD Helios Proof Point
Model selection and pod scheduling are mature. The layer between them — a policy-aware fabric deciding which inference site serves a request under latency, power, and sovereignty constraints — is AINF.
by Chris DePuy, 650 Group / September 24, 2026 This week, Meta, NEC, and Sumitomo Electric announced Petal, the first commercial petabit-class transoceanic submarine cable system, linking the US and France across roughly 7,000 km with service expected in 2029. Meta calls it the first multi-core fiber cable deployed at… Read More »Meta’s Petal Subsea Cable Doubles Transatlantic Capacity with Multi-Core Fiber
Networking now determines how effectively memory reaches GPUs Introduction Memory has become a critical constraint in AI infrastructure. As models scale, solutions involve a mixture of experts, and context windows expand, the way memory is used must change. If we don’t change, GPUs stall waiting for data rather than computing.… Read More »Marvell Expands AI Memory and Storage Infrastructure with Bravera SC6
This week at AMD Advancing AI Day in San Francisco (July 22–23, 2026), Arrcus and UfiSpace announced an integrated networking solution for AMD-powered AI infrastructure, shown alongside GIGABYTE’s GIGAPOD rack from Giga Computing built on AMD EPYC and AMD Instinct platforms. Arrcus is a leading networking software company; its ArcOS… Read More »Arrcus and UfiSpace Announce Ethernet System for AMD Instinct
