Vikram Sekar

Since I didn't actually to go to ECOC (epic fail, I should have), I had to live vicariously through announcements. There were still quite a few interesting ones related to optical scale up. Here are my thoughts on a few topics: 🕊️Teradyne Iris 100 for microLEDs 🕊️ams OSRAM thin-film VCSEL 🔒 Lumentum DWDM ELSFP 🔒 Lumentum + Qualcomm 1060nm VCSEL 🔒 Coherent Photonlink https://t.co/ugCKjw3OVK

Separating cooling of the NPO module from the switch ASIC opens up more opportunities as to where to place the laser source. https://t.co/S01tzgC6dz

"Anthropic is considering rolling out a new AI model to counter OpenAI’s momentum since its launch of GPT-6 Astra, according ​to three sources, ahead of an expected IPO and after its CEO called for an industrywide slowdown." The pacing AI argument aged like yesterday's milk left out in the hot sun. The AI buildout will continue. https://t.co/OsPErsM9kO

Where 4hi Actually Makes Sense: Edge AI "Edge AI players, not data center players, are the ones showing real interest in 4hi HBM. TrendForce observes that in certain autonomous driving and robotics applications, DRAM bandwidth has gradually become a performance bottleneck for computing platforms, prompting HBM to be included in architecture evaluations. There’s precedent for this. SK hynix supplied HBM2E to Waymo’s autonomous driving platform in 2024 and obtained automotive qualification, indicating that robotics and automotive applications have already become one of the extension areas for HBM usage. That said, these applications are still early. Relevant specifications remain under evaluation. As a result, actual demand volumes are limited, and the short-term contribution of these segments to overall HBM demand is expected to remain very low."

Vibe coded a system for me to keep track of my research with Codex Astra. It is integrated with Slack for interactions, Google Drive for Storage, Home Server for database, and OpenRouter for summarization. It took a good 4-5 hours of token burn, but nothing compared to what it would have taken to do manually.

So great to hear that CEOs of companies in photonics are reading my articles! My guess is they are referring to my posts on Lasers for CPO/NPO. I've been breaking down the physics of ultra high power lasers and what that means for the investment community as we start to first enter the world of NPO and then, later CPO. Lumentum has been the favorite choice for this, and there are good reasons why. But they are not without competition -- if not immediately, on the horizon. My job is to scan for those threats and report what I see from a physics-first approach. If you haven't had a chance to read those posts, here they are: Why CPO uses external lasers: https://t.co/qeGWOmjogG Lasers for CPO/NPO -- Part 1: The InP DFB Laser https://t.co/4fc82ymahT Lasers for CPO/NPO -- Part 2: Lumentum’s Technology and Moat https://t.co/JVWX70yybR Lasers for CPO/NPO -- Part 3: Master Oscillator, Power Amplifier https://t.co/9QJyfKDhSd

Uptime has been an afterthought in the rapid AI buildout but as AI becomes increasingly adopted by enterprise, reliability is going to become one of the biggest differentiating factors https://t.co/LrVvGpOIPL

Part 3 of the series on UHP lasers for CPO is out. In this post, we talk about the Master Oscillator Power Amplifier (MOPA) approach. Lumentum's approach has always been to make a single high power cavity that can produce about 400mW+ Others like Sumitomo do it differently... Their idea is to put a lower power CW DFB laser, and the put an amplifier after it to boost up to 400mW+ This approach has its difficulties and benefits, and the fact is you can make a really good laser by doing so. There are a lot of considerations that is covered in depth in the post. Finally, it looks at where Lumentum falls in all this and what to keep a watch for if you are a Lumentum investor. https://t.co/CIMd1rS3cq

Custom HBM can drive improvements in LLM inference by 5X. This is the case I made in my latest post, check out my pinned post. https://t.co/kWpyBW7H0r

Wrote up some thoughts on memory talks from Hot Chips today. We cover slides from SK Hynix, Micron and Samsung. Fully free post. Enjoy. https://t.co/6JcFkyJDWa