At Computex 2026 in Taipei, NVIDIA CEO Jensen Huang and Marvell CEO Matt Murphy emphasized that connectivity and networking technologies are becoming just as important as processors and memory in powering the next wave of artificial intelligence infrastructure. As AI models grow larger and data centers become increasingly distributed, high-speed interconnects are emerging as a critical component of AI performance.
Huang highlighted that the explosive growth of AI applications is driving unprecedented demand for infrastructure across the technology stack. He described Marvell as a potential “trillion-dollar company” because of its role in providing the networking and connectivity technologies that link AI accelerators, servers, storage systems, and data centers together.
Murphy noted that the future challenge for AI scaling is no longer limited to compute power alone. Instead, the industry’s next major bottleneck is moving enormous amounts of data efficiently between processors, memory systems, racks, and geographically distributed AI clusters. Marvell has invested heavily in optical networking, silicon photonics, switching technologies, and advanced interconnect solutions to address this challenge.
The discussion comes amid expanding collaboration between NVIDIA and Marvell, including integration through NVIDIA’s NVLink Fusion ecosystem and joint work on next-generation AI infrastructure technologies. The partnership reflects the industry’s growing focus on building complete AI factories rather than simply deploying standalone AI chips.
Why This Matters
• Highlights that networking and connectivity are becoming critical enablers of AI performance, alongside GPUs and memory
• Signals a shift from compute-centric AI infrastructure toward fully integrated AI factory architectures
• Strengthens the strategic importance of optical networking, silicon photonics, and high-speed interconnect technologies
• Supports the rapid expansion of hyperscale AI data centers worldwide
• Reinforces Marvell’s growing role in the global AI infrastructure ecosystem
• Demonstrates how AI demand is driving innovation across the entire semiconductor value chain
• Accelerates investment in next-generation networking, switching, and data-center technologies
• Shows that future AI growth will depend not only on more powerful chips but also on faster movement of data between systems and locations