Nvidia invests $4 billion in photonics companies Lumentum and Coherent
Nvidia announced Monday it is investing $2 billion each into photonics companies Lumentum and Coherent to develop optical transceivers, circuit switches, and lasers for next-generation AI data centers. The technology aims to improve energy efficiency, data transfer speeds, and bandwidth in data center infrastructure.
Nvidia announced Monday that it is investing $4 billion total—$2 billion each—into Lumentum and Coherent to accelerate development of photonics technology for data centers.
The investments fund development of optical transceivers, circuit switches, and lasers designed to move data at high speeds over long distances. According to Nvidia, the technology could improve energy efficiency, data transfer speeds, and bandwidth in future AI data center architectures.
For Lumentum, the deal is structured as a nonexclusive multiyear agreement that includes a multibillion-dollar purchase commitment from Nvidia. The Coherent partnership follows a similar structure.
These investments build on Nvidia's existing strategy to control data center infrastructure. The company acquired Mellanox in 2020 for network hardware capabilities, which it leveraged to develop NVLink technology that increases data movement between GPUs. Photonics represents the next layer of that strategy—moving data efficiently across entire data centers rather than just between individual processors.
Photonics technology has become critical infrastructure for large-scale AI operations. As AI models grow larger and training requires more distributed computing, the bottleneck increasingly shifts from compute power to data movement. Optical systems can transfer data faster and with lower power consumption than traditional electrical interconnects.
Nvidia's dual-investment approach reduces dependency on any single supplier while maintaining influence over technology roadmaps. Both Lumentum and Coherent are established players in optical networking, giving Nvidia leverage to shape products specifically for its data center ecosystem.
The timing aligns with accelerating demand for AI infrastructure. Major cloud providers and AI companies are building ever-larger data centers, creating urgent need for improved networking and power efficiency.
What this means: Nvidia is extending its control beyond chips into the complete data center stack—compute, memory, networking, and now photonics. This vertical integration creates higher barriers for competitors and locks customers into Nvidia's ecosystem for infrastructure decisions, not just GPUs. For Lumentum and Coherent, the investments provide capital and guaranteed demand but reduce their flexibility to serve competitors equally.
Related Articles
NVIDIA Kumo Relational predicts from databases using up to 10,000 in-context examples, no training
NVIDIA released Kumo Relational, a pre-trained foundation model that predicts outcomes such as churn, demand, and fraud directly on relational databases in a single forward pass. NVIDIA claims it beats supervised relational deep learning and expert-engineered gradient-boosted trees across 41 tasks on 15 databases, and surpasses AutoGluon ensembles by 8% MRR on SAP SALT.
NVIDIA releases Kumo Forecast, an Apache 2.0 time-series model built on a ~340M-parameter MOMENT-1-large encoder
NVIDIA has released Kumo Forecast (formerly NV-Tesseract Forecasting), a multivariate time-series forecasting model under Apache 2.0. It pairs a frozen MOMENT-1-large encoder of roughly 340 million parameters with a trainable forecasting head, optional nearest-neighbor retrieval over historical data (DARR), and a built-in interpretability framework.
NVIDIA opens RFD4-Proteina protein design model on Hugging Face Spaces, free dedicated GPUs until Nov 4
NVIDIA is sponsoring free access to a dedicated Hugging Face Space for RFD4-Proteina, an open BioNeMo protein design model co-developed with the University of Washington's Institute for Protein Design. Access runs from October 21 to November 4, 2026. NVIDIA claims the model produced 76.40 qualifying designs per GPU-hour at high thinking, versus 13.24 for the next-best method.
NVIDIA releases Nemotron 3.5 Embed 8B, scoring 61.17 nDCG@10 on ViDoRe v3.1 among dense retrievers
NVIDIA has released Nemotron 3.5 Embed (8B BF16, 1B BF16, 1B NVFP4), Nemotron 3.5 Rerank 8B BF16 and Nemotron 3.5 Parse (631M parameters) as open, commercially available models. NVIDIA reports that the 8B embedder ranks first among dense retrievers on ViDoRe v3.1 with 61.17 nDCG@10, and that adding the reranker cuts agent token cost per query by roughly 13-14%.
Comments
Loading...