가입 후 초대 링크를 공유하면 동영상 재생 및 초대 보상을 받을 수 있습니다.

Melvin
@MelvinInvests
AI Analyst @MilkRoadAI | Finding opportunities across AI, photonics, defense, space, and tech.
가입 June 2026
239 팔로잉 중    33.6K 팬
According to Korean media, Samsung is considering an acquisition to build up its silicon photonics business. Samsung Foundry plans to enter the photonic chip market next year on 12 inch wafers. Target customers include optical module makers like $COHR and $LITE. The same report says $TSM is still about 1 to 2 years ahead with its COUPE platform. Optical is turning into one of the biggest bottlenecks in AI data centers, and Samsung does not want to miss it.
더 보기
$NVDA just made some HUGE announcements about the future of AI infrastructure. At the Open Compute Project Summit, NVIDIA announced plans to open up more of its next generation Vera Rubin infrastructure designs while expanding its partnerships with $META, $GOOGL, and $CSCO. Essentially, NVIDIA wants to make it easier for companies to build massive AI data centers without having to design every piece of infrastructure from scratch. Its updated MGX architecture supports racks consuming up to 250kW of power, with standardized cooling, power connections, and modular hardware. This matters because as GPUs become more powerful, the infrastructure required to support them is becoming significantly more complicated. But I think the bigger story here is networking because Meta is adopting NVIDIA's next generation Spectrum 6 Ethernet switches, which offer 102.4 Tb/s of switching capacity. Combined with NVIDIA's networking architecture, these systems are designed to connect more than 100,000 GPUs while reducing the number of networking layers required. Think about what that means. We're reaching a point where AI clusters contain tens of thousands of GPUs that all need to communicate with each other extremely quickly. Having the fastest GPUs means very little if those GPUs are sitting around waiting for data. And this is exactly why I've been so bullish on the optical networking industry. As AI clusters continue scaling, we're going to need significantly more high speed optical connections, advanced networking equipment, and power efficient infrastructure. Companies involved in optical components, connectivity, and networking should benefit if these next generation designs translate into large scale deployments. That includes companies like $LITE, $COHR, $APH, and others across the supply chain, although the opportunity will vary by supplier. Bullish on AI networking. If you want to see how I’m investing around the companies powering the next generation of AI connectivity and infrastructure, check out my full Milk Road Pro portfolio below.
더 보기