Optical integration into switches (LPO, CPO) will surely be at the heart of DC networks in the coming years. Nvidia prepared for this partly through the Mellanox acquisition (2020), which now enables the integration of silicon-based optical circuits into Spectrum switches. This highlights Nokia’s role as a supplier of critical optical components (InP ELS), while Nokia simultaneously gains valuable experience for its own CPO development by mastering the entirety of SiPh and InP optics.
I have discussed the broader picture with Gemini, and the “in-depth analysis” suggests that this is a fundamental strategic shift. Nokia is not just tied to an AI-RAN breakthrough or a single DC architecture generation. Instead, it is building a holistic, AI-driven network development for current and future intelligent networks. Here are some highlights of this continuum:
Strategic Core: AI-Driven Vertical Integration
Nokia’s competitive advantage is not separate RAN, IP, and optical expertise, but the entire integrated stack optimized using artificial intelligence:
- 3 nm ASIC & DSP: Both routers (FP6) and optical modules (ICE7) are based on world-class 3 nm design expertise.
- InP & SiPh Foundries: Nokia controls the entire laser technology (InP) and silicon-based optics (SiPh) chain, ensuring a unique Western level of integration.
Digital Twin – Development Accelerator
Nvidia and Nokia are both industry leaders in utilizing digital twins. This collaboration makes them stronger together:
- Virtual Prototyping: Digital Twin enables the simulation of ASIC chips and optical networks in Nvidia’s Omniverse environment. This shortens development cycles by years and minimizes errors. This applies to RAN, IP, and optics alike.
- Network Automation (MantaRay & EDA): Digital Twin is not just a design tool; it drives network automation in real-time. For example, in telecommunications networks, Nokia’s MantaRay solutions extract even more capacity and reliability from networks (cf. AT&T automation). The same is done by EDA (Event-Driven Automation) in data center networks, enabling fully autonomous and error-free network management for AI workloads.
Behind the Smokescreen: AI-RAN vs. DC Infrastructure
While there has been much talk about AI-RAN in connection with the Nvidia agreement, it is only part of the whole. Justin Hotard’s silence on data center infrastructure (DC) in Q&A sessions suggests that Nokia is deeply integrated into Nvidia’s upcoming architectures (such as Rubin).
- AI-Driven Network Development: Nokia uses Nvidia’s AI power to develop next-generation communication technologies. This creates a continuum where AI designs the network, which is then optimized for AI traffic.
- Synergy in AI-RAN, MX Edge, and DC-DCI Networks: Even if, for example, AI-RAN does not make an immediate widespread breakthrough, Nokia can still leverage the development in the ASIC, radio, and transport design of different RAN versions. On the other hand, AI-RAN, MX Edge, and DC data centers represent very similar network architectures, so breakthroughs like 1.6T and 3 nm DSP support them all.
Investor’s Conclusion
Nokia is transforming from a cyclical hardware vendor into a critical enabler of AI infrastructure.
- 2026 Turning Point: Margins in H1/2026 will be tight due to investments, but the 1.6T launch and the ramp-up of InP factories pave the way for higher margins and a rerating of valuation multiples (P/E).
- Ecosystem Win: Once the “Nokia stack” is part of the Nvidia standard, it scales from data centers to AI-RAN and industrial edge computing (Edge).
This is not just a single agreement or deal, but a new way of operating: AI-driven network development in the AI era.