The scrutiny over the interrelationships of Nvidia, OpenAI, Oracle, CoreWeave, and other AI leaders continues, but the mammoth deals keep on coming. International Monetary Fund chief economist Pierre-Olivier Gourinchas weighed in to say that a bubble or a bust “won’t be a systemic crisis” like in the .com era, noting that cash-rich companies, as well as shareholders and equity holders, are the ones taking the risk this time around. No one wants to be left with costly assets, like excess capacity in data centers or stranded GPUs, so if AI companies buy chips from manufacturers, who in turn invest back into their customers, does it create an illusion of demand, or does it expedite much-needed compute and capacity innovation? The long-term outlook for AI infrastructure remains positive, with significant capital investments expected to create new markets and massive economic value. But with investor exuberance pushing valuations to unprecedented heights, some experts are warning it’s time for AI visionaries to get more picky about where to invest.

It’ll be interesting to see if more pragmatic, value-driven strategies and investments become the name of the game, as happened in the industrial IoT (IIoT) sector not long ago. Once characterized by hype and ad-hoc pilots, Industrial AI has moved IIoT from experimental pilots to boardroom strategies. Embedded in manufacturing strategy, governance, and performance frameworks, the global market for IIoT is now valued at $43.6 billion, with a 23% annual growth rate expected until 2030, at which point it’ll be worth $153.9 billion. Thanks to well-established use cases like automated optical inspection, machine vision and predictive maintenance, the foundation exists for transformative IIoT technologies that leverage generative AI and agentic AI for automation efforts and intelligent industrial systems.

Another sector eyeing generative AI and agentic AI is telecom, which is moving beyond initial experiments to implementation of these technologies for streamlining operations and delivering enhanced customer experiences. Bell Canada is leading the way with an initiative announced last week to leverage Cohere’s agentic AI platform North as part of its “Bell AI Fabric,” which will build “a sovereign, national AI infrastructure” in concert with the Canadian government’s broader $2 billion “Sovereign AI Compute Strategy.” Bell CEO Mirko Bibic told investors that he expects to hit $1.07 billion (USD) in AI revenue by 2028, in large part by leasing data center capacity, selling cybersecurity solutions, and integrating technology systems. To that end, Bell Canada will be building out a national network of data centers, six of which will be located in British Columbia (BC) for a total power capacity of 500 megawatts.