Today, at a Nokia event we attended, Hong Kong Broadband Network (HKBN) showed us how its 25G PON network can reach 20G upstream and downstream. Here is the screenshot. #25G PON

We’re excited to announce a strategic partnership with Fujitsu Global and 1Finity to deliver next-generation network solutions for the #AI era. As AI adoption accelerates worldwide, networks must scale seamlessly across data centers, clouds, 5G/6G, and edge environments.

AI is driving the future of security — and #DeepObservability is the key 🔑 to keeping up. Hear more from 650Group co-founders, @alanweckel and Chris DePuy | 650 Group, on the evolution of AI at #VisualyzeBootcamp: ow.ly/jQby50WNQRR

Here is my colleague Alan Weckel presenting at the @Nokia #SReXperts25 #Wavelengths25 conference today, speaking about #AI Infrastructure

At the Nokia Optical and Routing Conference, the big themes were clear: ⚡️ AI-driven traffic growth ⚡️ Rapid adoption of ZR coherent pluggables Our own Alan Weckel shared how this AI boom is reshaping network architectures — alongside insights from NTT DATA (Official), Zayo, Uber,

President of Arista Networks, Ken Duda sharing his views that the AVA service is critical to its customers. 650Group Alan Weckel

Arista Networks has filled out its branch strategy with the VeloCloud acquisition according to Arista Networks President Todd Nightingale Todd Nightingale 650Group Alan Weckel

I just published my latest blog on the 5G Americas conference — one of the most insightful wireless operator events of the year. Top themes we tracked this week included: 👉Satellite NTN/D2C, which has the potential to disrupt the mobile market more than any other topic at the

Just wrapped up ZTE Corporation User Conference in Milan — a week focused on AI, 6G, and operator realities in Europe. ZTE’s “Pre-6G” GigaMIMO trials with China Mobile hit 100 Gbps, 10× faster than 5G. AI, automation & competition were recurring themes from every operator. Full

Arguing that data center designs for AI must differ from traditional compute systems, Google VP, Partha Ranganathan shares some interesting growth stats at Google Google Cloud #OCPSummit25 Alan Weckel 650Group

Ian Buck of NVIDIA discusses a future 576 GPU per rack that uses 800V in the Kyber generation #OCPConference25 650Group Alan Weckel

Pradeep Vincent of Oracle Cloud Shared that the “speed of delivery” - how quickly Oracle can build a data center - leads to market leadership 650Group Alan Weckel #OCPConference25

.AMD EVP Mark Papermaster argues for open standards in the AI era and shares his company’s internal forecast for data center TAM. #OCPSummit25 650Group Alan Weckel

.Arm SVP Mohamed Awad pitched chiplets at #OCPSummit25 @alanweckel @650group

Ram Valega SVP Broadcom Argues that Ethernet is the only networking technology for AI data centers in no small part because the industry doesn’t want a single buyer using non-Ethernet. #ESUN #OCPSummit25 @alanweckel 650Group

DELL. SVP Ihab Tarazi In the past 2 months, 40 MW is the standard size of deployment these days and then these systems get connected together. Thanks @DELL for sharing these installation ratios! @alanweckel @650group #OCPSummit25

Similar to Oracle’s message yesterday, Denvr Dataworks, time to first token is everything and the company’s modular systems can accelerate it. According to them, you get 1MW in traditional DC in 6,400 sf while in its modular DC it only takes 600 sf. Alan Weckel 650Group

.@Intel announces Data Center GPU “Crescent Island” will be sampling in 2H26. #OCPSummit25 @alanweckel 650Group

.Meta VP Dan Rabinowitz says Meta needs bigger racks in part for power reasons which means Meta will continue using copper connectivity, thus there are two generations of Open Rack Wide (ORW). But, in future, Meta wants optically interconnected lower density racks, perhaps