Alan Weckel

At OCP Global Summit 2026, I sat down with Scintil Photonics for a three-part video series on optical networking and AI infrastructure. In Part 1, we discuss growing market demand and Scintil's approach to silicon photonics. Watch here: youtube.com/shorts/3xioeo7M3… #650Group #SiliconPhotonics #OCP2026

Scaling AI infrastructure is getting more complicated. Larger clusters, more demanding workloads and increasingly complex networks are putting pressure on how these systems are designed, deployed and managed. I recently discussed these challenges with Gilad Shainer, SVP of Networking at @nvidia, and Jason Forrester, VP of AI Cloud at IREN, including what operators are seeing in the field and how infrastructure requirements are evolving. Here's another clip from our conversation: youtube.com/shorts/Hosj1GpOO… #650Group #NVIDIA #IREN #AIInfrastructure

The AI infrastructure market is entering a new phase as investments in compute drive significant changes in networking architectures, data center design and infrastructure requirements. The scale of these deployments is creating new market opportunities and challenging traditional approaches to building and operating data centers. I recently hosted a discussion with Gilad Shainer, @nvidia and Jason Forrester, IREN to examine where the market is heading, the demands of building AI infrastructure at scale and the opportunities emerging as investment continues to accelerate. As AI deployments grow in scale and complexity, the decisions being made today will have significant implications for the infrastructure market over the next several years. Here's a preview of our conversation: youtube.com/shorts/Tkh68qusf… #650Group #NVIDIA #IREN #AIInfrastructure

At this month's @EthernetAllianc TEF 2026, I had the opportunity to discuss why 400G electrical technology is a critical next step in scaling AI networking and what it means for the future of Ethernet. As Ethernet expands beyond scale-out into scale-up AI architectures, were looking at a significant shift in both network design and market opportunity. At @650Group, we project Ethernet switching will exceed $300 billion this decade and potentially surpass $1 trillion in the early 2030s. But getting there will require more than faster technology. The industry needs to scale manufacturing and supply chains to support hundreds of millions of ports as AI infrastructure deployments accelerate. Thanks to @ConvergeDigest for the opportunity to share my perspective from TEF 2026. Watch the interview: lnkd.in/eTMnF6vM #AINetworking #Ethernet #400G #AIInfrastructure #650Group #TEF2026

The deep observability market grew 22% YoY in 1H 2026, with a projected 28% CAGR through 2030. Hybrid AI is driving demand for greater network visibility, security and performance. More on our latest @650Group research in my blog for @Gigamon: blog.gigamon.com/2026/10/15/…

Demand outran the venue at last week's OCP Global Summit 2026 in San Jose. Organizers reported roughly 15,000 attendees, with about 3,000 more turned away for space. My colleague here at @650Group, @ChrisDePuy tracked the release flow around the show: company releases timed to OCP went from nine in 2023 to 27 last year, and AI appeared in 49 of the 52 releases issued for this years show through day three. Networking made up roughly a third of them, up from 22% for all of 2025. 650group.com/blog/ocp-global… A few themes came up again and again: 🔹 Open rack-scale designs: More of the AI factory blueprint, from racks to compute trays, is being contributed to OCP. 🔹 Ethernet in the scale-up domain: Operators made the case that Ethernet can carry the accelerator-to-accelerator links once reserved for proprietary interconnects. 🔹 800 VDC and liquid cooling: Power and thermal design moved center stage as racks head toward megawatt scale. 🔹 Optics at every distance: Pluggable, near-package and co-packaged optics all had a seat at the table. What were your takeaways from OCP this year? Share them in the comments.👇 #OCPGlobalSummit #OpenCompute #AIInfrastructure #AINetworking #DataCenter #650Group

AI is changing the economics of connectivity, and the industry is only beginning to address what comes next. Great to participate in @SemtechCorp's Analyst Day and join the “AI Connectivity Imperative” panel alongside Mark Edelstone, David Ramku, Alexis Black Björlin and Dr. Hong Hou. From my perspective, networking and connectivity are becoming defining factors in how AI infrastructure scales. The discussion brought together some valuable perspectives on where the market is headed and what it will take to get there. Thanks to the Semtech team for including me and to Hong for sharing his executive perspective on the road ahead. #AI #Networking #Connectivity #650Group

102.4 Tbps switching is about more than bandwidth. Celestica new liquid-cooled system highlights a broader shift toward integrated networking, optics and cooling. As AI infrastructure scales, power efficiency and system-level design will be critical. themanufacturedata.com/new-1…

Great discussions yesterday at the OCP Global Summit, where I had the opportunity to participate in @OiForum’s workshop, Enabling an Open, Scalable, Sustainable AI Infrastructure. As AI infrastructure continues to scale, the industry faces significant challenges around bandwidth, power, latency and interoperability. Addressing these challenges requires collaboration across the ecosystem and a clear path from emerging technologies to real-world deployment. It was great to join industry leaders from across the OIF community to discuss the technologies shaping the next generation of AI networking, including 448G, coherent optics, energy-efficient interconnects and co-packaged optics. #OCPGlobalSummit #OIF #AINetworking #AIInfrastructure #OpticalNetworking #650Group

Co-packaged optics (CPO) continues to move from concept toward production, and @Celestica’s new DS6005 is another important step for AI networking. Built on Broadcom’s Tomahawk 6 – Davisson platform, the DS6005 delivers up to 102.4Tbps of switching capacity, combining CPO, liquid cooling and SONiC support in a system designed for high-density AI fabrics. As AI clusters continue to scale, power efficiency, bandwidth density, reach and reliability are becoming just as important as raw switching capacity. CPO has an important role to play in addressing these challenges, particularly in scale-out networking. As I noted in Celestica’s announcement, the opportunity for CPO is moving beyond the technology itself to how it can be deployed reliably at scale. Collaboration across silicon, optics, systems and manufacturing is critical to making that transition. Read the announcement: appeal-democrat.marketminute… #AIInfrastructure #CoPackagedOptics #Networking #OCP #650Group

Tomorrow at @OpenComputePrj OCP Global Summit, I’ll be joining @OiForum and a great group of industry experts for a half-day discussion on what it will take to build open, scalable AI infrastructure. We’ll cover everything from 448G electrical interfaces and energy-efficient interconnects to NPO, CPO, coherent optics and multi-vendor interoperability. If you’re at OCP, join us: OIF Workshop: Enabling an Open, Scalable, Sustainable AI Infrastructure Tuesday, Oct. 13 1–5 p.m. PT San Jose Convention Center, Room LL20BCD Hope to see you there! #OCPGlobalSummit #AIInfrastructure #OIF

AI networking is scaling rapidly, but the manufacturing supply chain for optical interconnects is not keeping pace. In my latest @650Group blog, I look at one of the less-discussed constraints on AI infrastructure growth: indium phosphide (InP) and the industry's ability to manufacture optical components at the volumes AI factories will require. AI networking is on a path to exceed $300B this decade and approach $1T early next decade. But getting there will require more than faster switches, new architectures and open standards. We need to rethink how optical components are manufactured, moving from telecom-era production models to semiconductor-scale manufacturing capable of supporting billions of lasers. The industry made major investments in advanced packaging and HBM to keep pace with GPU demand. Networking needs a similar shift. I explore why heterogeneous integration of silicon photonics and III-V materials will be critical to scaling AI infrastructure, and why manufacturing must advance alongside networking innovation. Read the full blog: 650group.com/blog/ai-buildou… #AIInfrastructure #AINetworking #OpticalNetworking #SiliconPhotonics #650Group

Next week, I’ll be joining @OiForum at the 2026 @OpenComputePrj OCP Global Summit for a half-day workshop focused on the technologies needed to scale AI infrastructure. The industry is moving quickly from individual technology discussions to a much bigger systems question: how do electrical interfaces, optics, power efficiency and interoperability come together as AI infrastructure scales? We’ll be digging into 448G, NPO and CPO, coherent optics, energy-efficient interconnects and multi-vendor interoperability with experts from across the ecosystem. **OIF Workshop: Enabling an Open, Scalable, Sustainable AI Infrastructure** Tuesday, Oct. 13 | 1–5 p.m. PT San Jose Convention Center Looking forward to the discussion. #OCPGlobalSummit #AIInfrastructure #OpticalNetworking #Interoperability

Heading into #OCPGlobalSummit, open networking and SONiC will be a big part of the AI infrastructure conversation. I shared my perspective in @Cisco's blog on where scale-across is heading and why we expect this market to exceed $50B by the end of the decade. Worth a read: blogs.cisco.com/datacenter/s… #SONiC #AINetworking #650Group

Today at @EthernetAllianc #TEF2026, I’m moderating two discussions on whether optical infrastructure can scale with AI: 1:45 p.m. PT — Optical infrastructure challenges 3:45 p.m. PT — Multi-core fiber AI can’t scale if the physical infrastructure can’t scale with it. ethernetalliance.org/tef-202… #TEF2026

From @MarvellTech's Investor Day in NYC: the scale of the AI infrastructure opportunity continues to expand. Marvell sees TAM reaching $400B by 2030, supported by nearly $3T in CAPEX. At 650 Group, we see networking as a critical enabler across Scale-In, Scale-Up, Scale-Out and Scale-Across — with AI networking exceeding $1T in the early 2030s. My latest blog: Marvell Investor Day – The Path to Multi-Trillion CAPEX and Tokenomics 650group.com/blog/marvell-in… #AI #AINetworking #DataCenter #Marvell

Heading to San Jose for @OpenComputePrj #OCPSummit26? I’ll be there Oct. 12–15 and would welcome time with companies working across switching, optics, silicon, systems, power and cooling. If you’d like to compare notes, send me a message. opencompute.org/summit/globa…

Looking forward to joining @OiForum at the 2026 @OpenComputePrj OCP Global Summit in San Jose for its half-day workshop focused on the infrastructure behind AI at scale. I’ll be participating alongside a great group of industry experts from Microsoft, Cisco, Broadcom, Oracle, TE Connectivity, Ranovus, Amphenol, Ciena, Nokia, HPE and Huawei to dig into the technologies that will be critical as AI infrastructure continues to scale. OIF Workshop: Enabling an Open, Scalable, Sustainable AI Infrastructure 📅 Tuesday, Oct. 13 ⏰ 1–5 p.m. PT 📍 San Jose Convention Center, Room LL20BCD Looking forward to the discussion and to seeing many of you at OCP Global Summit. #OCPGlobalSummit #AIInfrastructure #OpticalNetworking #Interoperability

The AI infrastructure build-out is only part of the equation. The next challenge is operating it efficiently. I joined @nvidia's Gilad Shainer and @AvizNews’ Vishal Shukla to discuss AI factory networking, automation, multi-tenancy and the operational layer behind GPU utilization. Watch: lnkd.in/eV6Un793 #AINetworking #AIFactory

As AI infrastructure scales, the SONiC conversation is moving beyond openness and choice to a harder question: how do you operate it successfully at AI scale? I’ll moderate @HPE's Reimagining SONiC to Power AI at Scale webinar on Oct. 6. Register: rsvp.hpe.com/reimaginingsoni… #SONiC #AIInfrastructure #Networking #HPE