Nvidia is preparing to open the door to photonic silicon . Just a few hours ago, it has started in Palo Alto, California (USA), the Specialized Conference in Semiconductor Engineering and High Performance Computing known as ‘Hot Chips’. The company, led by Jensen Huang, has seized this opportunity to announce that in 2026, its latest generation artificial intelligence (AI) platforms will utilize photonic interconnections to achieve higher transfer speeds between GPU clusters.
Most designers and manufacturers of integrated circuits are currently focused on developing silicon photonics . Douglas Yu, a TSMC executive responsible for systems integration, articulated its disruptive potential in September 2023: “If we manage to implement a good system of integration of silicon photonics, we will trigger a new paradigm. We will probably place ourselves at the beginning of a new era .”
Nvidia’s Photonic Silicon Ambitions
Before delving deeper, it’s essential to understand what we mean by photonic silicon . This emerging discipline seeks to harness the properties of silicon to optimize the transformation of electrical signals into light pulses . The primary application of this technology lies in establishing high-performance links that can enhance communications between multiple chips and facilitate the transfer of information between various machines.
Advanced packaging technologies employed by leading semiconductor manufacturers such as TSMC, Intel, and Samsung can significantly benefit from efficient communication mechanisms between high-performance chips. Moreover, large data centers , where a multitude of machines must be interconnected, stand to gain immensely from this innovation. However, the most promising application of photonic silicon occurs within the realm of artificial intelligence .
CPO technology reduces energy consumption to just 9 watts per port
This is precisely Nvidia’s ambition. In these clusters, thousands of GPUs must work in unison; thus, establishing high-performance links is critical. Traditional solutions like copper cables or optical modules can address this issue, but they introduce substantial inefficiencies, particularly in terms of energy loss and potential bottlenecks. The data transfer can consume up to 30 watts per port , resulting in excessive heat generation and increased failure rates.
Furthermore, latency poses a limitation to the scalability of clusters as the number of GPUs in data centers climbs. To mitigate these inefficiencies, Nvidia plans to integrate optical components that necessitate photonic interconnections directly within the encapsulated switching chip . This innovative technology, referred to as CPO (Co-Packaged Optics) , will reduce energy consumption to just 9 watts per port , minimize signal loss, and enhance data integrity. Early indications suggest this technology holds significant promise.
Nvidia has confirmed that it will incorporate CPO technology into its Quantum-X InfiniBand and Spectrum-X Ethernet interconnection platforms by 2026. Importantly, CPO will not be an optional enhancement; it will be integrated as a fundamental requirement for the next generation of AI data centers , aiming to bolster the competitiveness of Nvidia’s hardware platforms.
Image | Nvidia
Further information can be found in the article from Tom’s Hardware.
As the race for AI capabilities intensifies, advancements in silicon photonics not only pave the way for improved efficiency and performance but also signify a pivotal shift in the underlying architecture of computer systems. These technologies will enable data centers to handle increasingly complex applications while significantly reducing energy consumption and enhancing data handling capabilities. The implications span various industries, promising a future where high-speed data transfer becomes the norm rather than the exception. Nvidia’s commitment to integrating photonic technologies underscores the importance of innovation in maintaining a competitive edge in the fast-evolving tech landscape.