Over the past few years, I have written several analyses on the profound structural changes unfolding across the ICT sector. I have discussed how artificial intelligence is transforming data centres into vast energy consumers, how telecommunications networks are shifting from “dumb pipes” to intelligent systems, and how digital technologies increasingly intersect with energy policy, governance, and sovereignty.
At the time, these trends appeared to run in parallel — significant but still loosely connected. However, the extraordinary developments from Nvidia and OpenAI over the past months make clear how all these threads are now converging. Together, they reveal the outline of a new industrial landscape: one defined not by physical or digital infrastructure alone, but by cognitive infrastructure — an environment where computing, communication, and cognition form a single, interdependent system.
The industrialisation of AI
In recent weeks, Nvidia has announced a series of partnerships that together map out the next phase of the digital economy. The company revealed collaborations with Eli Lilly on AI-driven pharmaceutical research, Hyundai on robotics and autonomous manufacturing, Uber on a 100,000-vehicle robotaxi project, and Samsung on semiconductor design.
Equally significant are its alliances with the U.S. Department of Energy to build seven new supercomputers, Deutsche Telekom to integrate AI into network management, and Nokia, where a US$1 billion investment will embed AI into the operation of 6G networks and prepare for AI-enhanced workstations and smartphones. Nvidia also unveiled a new system linking its AI chips directly to quantum computers.
These are not isolated deals. They represent a coordinated industrial strategy — extending AI from data centres into networks, devices, vehicles, and scientific research. Nvidia is positioning itself as the backbone of a new cognitive economy.
The Nvidia–OpenAI axis
Equally pivotal is the deepening relationship between Nvidia and OpenAI. Their collaboration dates back to 2016, when Nvidia delivered its first DGX-1 supercomputer to OpenAI. Since then, Nvidia’s hardware has powered every generation of OpenAI’s large-language models, including GPT-4 and beyond.
Recent announcements show this partnership scaling to new levels. Nvidia plans to invest up to US$100 billion in OpenAI infrastructure, deploying as much as 10 gigawatts of computing capacity. This is more than a commercial alliance; it marks the fusion of the world’s leading AI software developer and its dominant hardware supplier — a powerful axis at the centre of the global technology system.
It confirms what I argued in earlier writings: computing capacity itself has become a form of strategic infrastructure, comparable to energy, transport, or communications. Access to compute will define economic and political power in the coming decades.
Compute as the new critical infrastructure
The industrial age ran on coal and oil; the digital age runs on data and compute. Today’s AI build-outs demand immense energy, land, and materials — placing them squarely in the domain of national infrastructure. Governments and corporations alike are now competing to secure chips, energy supply, and data pipelines.
This new dependency challenges existing policy frameworks. Nations used to plan for electricity and transport; now they must also plan for computing capacity. The geopolitical competition over semiconductors and the scramble to build AI-ready data centres are early signs of this new industrial reality.
The merger of AI and telecommunications
In previous analyses I have argued that telecommunications networks must evolve from passive conduits to active participants in the intelligence economy. That argument is now being validated in real time.
Nvidia’s investment in Nokia and its partnership with Deutsche Telekom signal that the network itself is becoming an intelligent organism. AI will increasingly manage and optimise bandwidth, latency, and energy use, adapting in real time to shifting demand.
The boundary between computing and communication is disappearing. Networks are no longer just highways for data — they are becoming cognitive systems, learning from the traffic they carry. This echoes the transition I described in my article Telstra’s bold reinvention: breaking free from its dumb pipes past, where I argued that telecom operators must reinvent themselves as intelligent infrastructure providers.
The decentralisation of intelligence
At the same time, AI is spreading outward — from centralised data centres to the edge of the network. The Nokia partnership is a preview of what is to come: AI embedded not just in the cloud but in workstations, mobile devices, and sensors.
These distributed systems — what have been referred to as Personal AI or #PAI — will cooperate and communicate much like early PCs once did through local networks and the cloud. The result will be swarms of interconnected AI agents capable of learning, adapting, and collaborating.
This new topology of intelligence will generate unprecedented data flows and complexity, demanding continuous upgrades in communication infrastructure and energy capacity. It also validates what I discussed in From power grids to people, where I argued that digital and energy systems are converging into one interdependent ecosystem.
The energy dimension
The scale of AI expansion brings with it a fundamental challenge: energy. Training and deploying AI models consumes enormous amounts of electricity and cooling. Nvidia’s own infrastructure plans — and those of its partners — will add several gigawatts of demand globally.
This blurring of digital and physical infrastructure is not hypothetical. AI now shapes patterns of energy production, consumption, and investment. The U.S. Department of Energy partnership points to a new model where computing becomes a component of the energy grid itself.
This is precisely the scenario I warned about earlier — that data-centre growth and AI deployment could reshape global energy systems, turning electricity grids into the nervous system of the digital economy.
Cognitive infrastructure: the new architecture of progress
All these developments point toward the emergence of what I call cognitive infrastructure — a system in which intelligence is built into every layer of the economy.
It integrates computation, communication, and energy into a single fabric that can sense, analyse, and act. This infrastructure will no longer merely support economic activity; it will increasingly decide how that activity unfolds.
In practical terms, this means AI will design logistics, allocate resources, optimise energy grids, manage networks, and eventually coordinate other AIs. The infrastructure of the future will not just process information — it will think (in the artificial sense).
Governance and opportunity
This transformation brings immense opportunity but also new dependencies. The concentration of AI infrastructure among a few global players raises questions of access, sovereignty, and resilience.
For businesses and policymakers, the challenge will be to navigate this shift from digital to cognitive infrastructure without repeating the centralisation mistakes of the past. The winners will be those who recognise that intelligence is now part of the infrastructure stack — and plan accordingly.
We must therefore start thinking beyond broadband plans and data strategies, toward comprehensive frameworks that treat AI, energy, and communications as one system. The goal is not to slow the rise of cognitive infrastructure, but to ensure it serves society rather than the narrow interests of a few technology conglomerates.
The next phase
The rise of cognitive infrastructure represents the culmination of decades of technological evolution. It connects every major theme I have explored in my previous writing: the transformation of telecoms, the energy cost of AI, the restructuring of global supply chains, and the risks of digital dependency.
What we are witnessing is the moment when technology stops being a set of tools and becomes the architecture of cognition itself — the infrastructure that will define the next industrial era.
The question, as always, is whether we will build it with wisdom.
Paul Budde