The AIโ Boom Drives Innovation in Chip Networking
The โฃsurgeโ in demandโฃ for artificial intelligence is pushing the boundaries of data transfer โฃtechnology, as conventional electronic interconnects struggleโ to keep pace withโข the bandwidth requirements of modern AIโ workloads.This has sparked โฃa wave of innovation focused on accelerating data throughput within โand between computer systems.
Historically, โnetworking technology was a relatively stable field, primarily concerned with efficiently switching packetsโค of data. However, the computationally intensive nature โof AI has dramatically altered this landscape. “Now, becauseโข of AI, it’s having to move fairly robust workloads, and that’s why you’reโข seeing innovationโข around speed,” explains Ben Bajarin, CEO โofโข the research firm Creative Strategies.
Nvidia recognized this โshift early on, making strategic acquisitions to bolster its networking capabilities.โค In 2020, the company acquired Mellanox โฃTechnologies for nearly $7 โคbillion, gaining access to high-speed networking solutions for servers and data centers. Shortly after, Nvidia also purchased Cumulus Networks, enhancing its Linux-based softwareโข system forโข computer networking, betting that clusteredโ GPUs โคwould become considerably โฃmore powerful within data center environments.
While Nvidia focusesโข on โฃvertically integrated โคGPU โคsystems, Broadcom has โฃemerged asโข a key player inโค custom chipโ accelerators and high-speedโฃ networking. The โคcompany collaborates withโ major players like Google,โ Meta,โ and OpenAI, โdevelopingโข chips specifically for data centers. Broadcom is also โคa leader in silicon photonics and โคis preparing to launch the Thor Ultra networking chip,designed to optimize data flow between AI systems and the broader data center infrastructure,as reported by Reuters.
Further demonstratingโ the industry’sโ focus,ARM,a semiconductor design giant,recently announced its acquisition โof DreamBig for $265 million.DreamBig specializesโ in AI chiplets – modular circuits designed for integration into larger systems – in partnership with Samsung. ARM CEO Rene Haas highlighted the acquisition’sโฃ importance for both “scale-up and scale-out networking,” โreferring to efficient data transfer within and between chip clusters โฃand racks.
A especiallyโ innovative approach โis being pioneered by Lightmatter,โ which is developing silicon photonics to โขlinkโ chips โขtogether. CEO Nick Harris notes that AI’s computing power demands are doubling every โคthree months, exceeding the pace predicted byโข Moore’s Law. Lightmatter’s technology utilizes light-based interconnects,creating a 3D stack of silicon that theโ company claimsโ represents the “world’s fastest photonic engine for AI โchips.”โ Theโข startup has secured over $500 million in funding, reaching a valuation of $4.4 billion in the past two years, reflecting investor confidence in this novel approach to โovercoming data transfer bottlenecks.