The hidden cost of the digital revolution
According to the IEA, the largest data centre currently under construction will need as much energy as two million homes. In the United States, nearly half of electricity demand growth through 2030 will come directly from AI, already surpassing the consumption of energy-intensive industries like steel, aluminium or cement. There’s an even more telling detail about this technology’s nature: during training of the most advanced models, electricity demand can spike by 400% within a fraction of a second, forcing entire grids to be redesigned with battery storage to absorb those swings.
The takeaway isn’t that AI is inherently unsustainable — it’s that its sustainability depends entirely on how well the data behind it is managed: generation data, consumption data, grid data. Without that data layer properly instrumented, any digital infrastructure risks becoming an energy problem before it becomes a solution.
When data becomes industrial efficiency
This is where the second report of the week comes in. KPMG finds that 76% of manufacturing companies already show strong readiness to adopt advanced technologies, and 34% are already seeing ROI across multiple AI use cases: predictive maintenance, machine-vision quality control, process optimisation. In Spain, projects developed by Telefónica and its industrial engineering subsidiary Geprom for companies like Cerealto, Deoleo and Legumbres La Cochura show this transformation is no longer a future promise but a reality already running on food and logistics production lines.
The IEA confirms it from the other end of the equation: if AI applications for grid optimisation, demand forecasting and renewables integration were deployed at scale, the potential energy savings by 2035 could equal the annual consumption of Indonesia — a country of nearly 290 million people. The technology that consumes can also give back.
Engineering as the hinge between consumption and efficiency
The common thread running through both reports is the same one we keep coming back to in this newsletter: without reliable, well-captured, well-managed data, AI isn’t sustainable and industry isn’t efficient. The difference between a factory that cuts consumption by 20% and a data centre that spikes a country’s electricity demand isn’t the technology itself — it’s the quality of the engineering behind it: well-calibrated sensors, robust data architecture, and an energy strategy built into the design from day one, not bolted on afterwards.
At Talat, we believe that’s exactly engineering’s role this decade: not choosing between innovation and sustainability, but designing the systems that make sure one can’t exist without the other.
How is your organisation making sure its digital projects add efficiency rather than subtract it? We’d love to hear your take in the comments.