Justin Currlin

Proud to share that I was recognized as Salesperson of the Month for the Mobile Computing Division at KEYENCE CORPORATION for May. I finished the month at 573% to goal and closed out Q1 at 258% to goal. Huge thank you to Brent Hopkins for your guidance and support. It’s been a rewarding year so far, and I’m excited for what’s ahead.

Jay Kirkland just left AMD — where he led EPYC and Instinct platform engineering — to become Marvell's SVP of custom silicon. On the way out he said the AI opportunities in front of him are "expanding faster than our industry's ability to deliver them." Read that again. That's not a demand problem or a capital problem. It's a people problem, said out loud by one of the few people qualified to fix it. Here's what I'm watching from the search desk. Kirkland is the second senior hire Marvell has landed in weeks — Dan Durn came in as CFO from Adobe in June. When a company whose stock is up 200%+ starts stacking proven operators this fast, it's telling you where the real bottleneck is: not chips, but the tiny pool of leaders who've actually shipped silicon at scale. That list is basically Intel, AMD, Broadcom and Nvidia alumni — and every AI-infrastructure and networking-silicon company is fishing the same pond. The companies winning right now aren't the ones with the biggest roadmaps. They're the ones who can pull a Kirkland before a competitor does. If you build AI networking or custom silicon: is your top engineering talent a retention plan, or a resignation waiting to happen? I run executive search in networking, telecom, and AI infrastructure, and I post on where the talent is actually moving. Follow along if that's your world. Link in comments.

We're coming out of stealth. We've built our first racks after a successful A0 tapeout, $1B+ in customer contracts, and $800m raised. Early customer tests show us achieving SOTA throughput, latency, and power efficiency on inference workloads. Our first racks ship this summer. We're a team of 400+ engineers from NVIDIA, Google TPUs, Broadcom, SK Hynix, TSMC, & more. We're backed by Jane Street, HRT, Two Sigma, and Jump, with strategic investment from VentureTech Alliance. We're excited to deepen our partnership with the world's leading semiconductor manufacturer. Our Series B was led by Stripes, with participation from Ribbit Capital, Radical Ventures, Positive Sum, Primary, & Argo. Our inference systems are built to push the entire pareto curve on frontier models, including many-trillion parameter MoEs, long context, and agentic workloads. Today, we're sharing two breakthroughs to make this happen: Low-Voltage Inference (LVI) for high throughput workloads. Today, AI chips can't scale FLOPs without thermal throttling. As FLOPs utilization increases, AI chips draw more power and downregulate clock speed. This often results in sustained inference throughput under half of peak FLOPs. Chips in other industries solve the power problem by running at lower voltages. Bitcoin miners run at under 3x the voltage of AI chips! We’ve designed a new architecture to run our chip’s math blocks at under half the voltage of most AI chips. This enables multiple times the FLOPs density of AI chips today. Cluster-Scale Memory (CSM) for low latency workloads. Today's AI chips using HBM can’t achieve SRAM-level decode speeds due to memory subsystem and interconnect bottlenecks. SRAM-only chips have lower FLOPs density and memory capacity, sacrificing throughput. You’re forced to make a tradeoff: serve at much slower speeds, or run at low batch sizes and suffer from higher costs. When running large MoE models, token routing across experts requires sending data through a deep memory hierarchy and a networking switch to reach a destination expert. Each memory layer inherently adds latency; thus, the best layer is no layer. We’ve designed a new architecture that creates a shared low-latency memory pool across the entire scale-up domain. We use a proprietary ultra-low-latency, high-bandwidth interconnect to enable dramatically faster memory access across chips. Our HBM/SRAM hybrid design solves both memory capacity and mem2mem latency, enabling high throughput and interactivity simultaneously. CSM improves latency and avoids today's cost, reliability, yield, thermal, and compute tradeoffs of SRAM-only chips, 3D DRAM chips, or optics. We're scaling production as fast as possible. We built a 2MW datacenter in our office and opened a Taiwan factory for 24/7 engineering. Performance, roadmap, and more updates are coming this summer. If this excites you, join us to build the future of gigawatt-scale inference: https://lnkd.in/gaQBX-Fz.

Big day at Baseten! $1.5 Billion Series F 20x Revenue 40x Inference Volume Inference isn't just growing — it's becoming one of the most important markets ever created. So excited to be a part of this team!

I hate to admit how long into my leadership journey it took me truly comprehend this critical point. Not all that long ago, I found myself standing in front of a whiteboard after a particularly frustrating meeting. We had spent the better part of an hour discussing performance, opportunities, priorities, and next steps - and I needed to illustrate to myself my main takeaways and relative action items. In the meeting, everyone agreed. Everyone nodded while on mute. But after a few quick chats with trusted individuals on the side, it was abundantly clear we all left with a different understanding of what we had just been decided. That, or we legit didn’t know and didn’t want anyone else in the call to know we were so lost that we didn’t even know where to begin. That was the moment I realized alignment isn’t measured by agreement, rather it’s measured by whether people can leave the same room and explain the outcomes in the same way. Since then, I’ve become a lot less interested in whether everyone agrees during a discussion and a lot more interested in whether expectations are clear when the discussion ends. Back briefs are a wonderful tool to help master the check - and can be implemented in a variety of ways using MS365, Slack and any other major platform. If you haven’t incorporated a process for confirming your message landed, start simple and start today :)

The best decisions rarely feel clean in the moment, there's almost always a tradeoff. Investing in one area means not investing in another, moving faster means accepting risk, waiting for more data could mean missing your window. I've spent a lot of time trying to find the “right” answer - the one that checked every box. Over time I’ve learned that most meaningful decisions don’t work that way. They require clarity on priorities, alignment on what matters most and a willingness to move forward without perfect information. Progress tends to follow those strong enough to make those calls & then own the results.

Introducing the VS‑G, KEYENCE’s most powerful vision system to date. Engineered for high‑speed, heavy‑duty inspection, the VS‑G delivers breakthrough processing performance, intelligent image storage, and automated parameter optimization — all designed to make complex applications easier, faster, and more reliable. If you’re pushing the limits of what machine vision can do on the manufacturing floor, the VS‑G sets a new benchmark. To learn more about the NEW VS-G click here: www.keyence.com/VS-G #KEYENCE #VSGseries #MachineVision #AI #NewInVision #MachineLearning #VisionSystem