\n<\/p>\n
At the Data Center Knowledge News Desk during <\/span>Data Center World 2025<\/a><\/span>, Omdia\u2019s Manoj Sukumaran shared his perspective on the AI computing revolution, comparing it to the evolution of mobile networks.<\/span><\/p>\n \u201cWe\u2019re in the 2G era of AI,\u201d he explained, suggesting the industry is only at the beginning of a transformative journey.<\/span><\/p>\n Addressing the projected <\/span>$1 trillion in global data center capex by 2030<\/a><\/span>, Sukumaran emphasized that current computing systems aren\u2019t optimized for large language models, necessitating a complete infrastructure overhaul. He pointed to solutions emerging across the industry, including novel approaches to power generation \u2013 such as ship engines potentially being adapted for data center use.<\/span><\/p>\n Sukumaran highlighted the critical challenge of data movement, which currently consumes 60% of energy in AI systems. He detailed how next-generation chips, like Nvidia’s future <\/span>Rubin Ultra<\/a><\/span>, are tackling this through the integration of compute and memory elements in single packages.<\/span><\/p>\n The analyst maintains that while the industry faces significant challenges in power, cooling, and infrastructure, the combination of innovation and resources will drive solutions forward as AI computing scales from today\u2019s 100,000-GPU clusters to anticipated million-scale deployments.<\/span><\/p>\n \u25b6\ufe0f <\/span>Watch the full interview here:<\/span><\/span><\/p>\n