Artificial intelligence (AI) adoption is accelerating faster than infrastructure readiness, according to the Beyond the Bottleneck: AI Cluster Networking Report 2025 by Keysight Technologies, Inc., in collaboration with Heavy Reading.
Keeping up with AI adoption
The report found that 62% of respondents stated they aim to maximise the benefits of their existing infrastructure without requiring new investment.
Moreover, almost all operators (95%) believe that real-world AI workload emulation can help validate performance, enhance efficiency and deploy more next-gen AI clusters.
Most respondents (89%) plan to expand or maintain their AI infrastructure investments in the coming year, with cloud integration (51%), faster GPUs (49%), and high-speed network upgrades (45%) driving the growth of their investments.
However, despite the challenges, 50% of operators are determined to overcome budget constraints (59%), infrastructure limitations (55%), and talent shortages (51%) to scale AI.
Some 34% are exploring 800G, 22% are trialling 1.6T, and a significant 58% are evaluating Ultra Ethernet as a high-performance networking option, indicating the rise in adoption of advanced networking technologies. The growing interest in 1.6 T networking is a clear sign of the industry's evolution. Network capacity is also emerging as a decisive factor in AI scalability, with 55% of operators deploying 400G interconnects.
"AI data centres are reaching a tipping point where performance and scale alone are not enough. Operators need deeper insight, tighter validation, and smarter infrastructure choices," said Ram Periakaruppan, vice president and general manager, Network Applications & Security Group at Keysight. "This research confirms what we see in the field: success in the AI era hinges on optimising every layer of the network."
