The Dashboard Nobody Redesigned
Central bankers have spent decades learning to read a specific set of instruments. Capacity utilization tells them whether factories are running hot. Inflation data tells them whether demand is outpacing supply. Employment numbers tell them whether the economy is running near its limits. These gauges were calibrated over generations, refined through recessions and recoveries, and trusted precisely because their movements had been observed long enough to be understood.
That trust is now being tested by something the frameworks were not built to read.
A study published by the Bank for International Settlements on October 8, 2026, argues that artificial intelligence’s environmental footprint — specifically its ravenous electricity consumption — is blurring the readings on that dashboard. [1] The paper, BIS Paper No. 174, was written by economists Leonardo Gambacorta and Salvatore Polizzi. [1] Its central claim is not that AI is bad for the economy. It is that AI is making the economy harder to read.
The distinction matters: the paper is not a verdict on AI, it is a warning about measurement.
What the Paper Actually Found
The BIS paper does not treat AI as a monolithic force. It credits the technology with genuine contributions to climate mitigation and adaptation — improvements in energy efficiency, more accurate forecasting, and innovation in low-carbon technologies. These are not trivial gains. Better forecasting alone can reduce waste across energy systems and supply chains.
But the same systems that deliver those gains require enormous amounts of power to operate. Data centers housing AI models add significantly to electricity consumption and the emissions that come with it. The International Energy Agency projects that global data center consumption could exceed 945 terawatt hours by 2030 — more than double current levels. That is a significant new load on energy systems that were not designed with this demand in mind.
The paper notes that this energy intensity puts upward pressure on electricity prices and contributes to grid constraints. Those are economic effects, not just environmental ones. When electricity prices rise because data centers are bidding for power, that increase shows up in the same inflation data central banks use to set interest rates. The question the BIS authors raise is whether policymakers can distinguish between price increases driven by excess demand and price increases driven by data centers bidding up electricity.
Misreading one for the other has consequences for rate decisions.
The Signal and the Noise
Consider the puzzle the paper lays out. Output is climbing. Is that because the economy is overheating, or because AI has made workers more productive? Prices are rising. Is that because demand exceeds supply across the board, or because data centers are driving up electricity costs in specific regions?
A central bank that mistakes an energy-driven price increase for broad overheating could push toward a policy move the economy does not need.
The BIS paper argues that AI-driven changes in productivity and energy constraints could muddy exactly these signals. Capacity utilization, a key indicator of whether the economy is running near its productive limits, becomes harder to interpret when AI is simultaneously changing what those limits are and consuming resources that might otherwise be available to other sectors.
This is the structural problem the paper describes: the gauges are smudging.
The Investment Question
There is a second dimension to the BIS analysis that extends beyond monetary policy. The paper highlights the substantial investments flowing into AI infrastructure and energy generation.
The concern is not that the technology fails, but that the investments deliver disappointing returns.
The paper does not quantify the risk. It notes that the scale of investment makes it worth monitoring, and that disappointing returns could pose risks to financial stability.
Two Roads, Neither Fully Mapped
The first is incremental, with AI gradually woven into markets and the wider economy. The second is transformative: a leap toward artificial general intelligence, or AGI, the hypothetical point where AI matches broad human capability.
Each scenario carries its own uncertainties for economic outcomes and policy implications. An incremental path gives central banks time to adapt their frameworks, to develop new indicators that separate AI-driven productivity gains from traditional demand signals. A transformative path compresses that timeline. If AGI arrives suddenly, the economic disruption could be severe enough that traditional monetary policy tools become inadequate or misleading.
They simply note that both are possible and that policy frameworks need to account for the uncertainty.
The Institutional Dimension
What makes the BIS paper notable is not any single finding. It is the institutional context in which those findings appear. The Bank for International Settlements is not a think tank or an academic journal. It is the central bank of central banks, an institution whose publications are read by the people who actually set interest rates.
When the BIS says that AI’s energy footprint could obscure the indicators policymakers rely on, it is flagging a practical problem for the institutions that manage the global economy. The paper’s core message is that environmental effects belong inside traditional monetary policy frameworks — that treating AI’s energy demand as a separate climate topic risks missing how it shows up in inflation and output data.
The BIS paper argues that AI has collapsed the distinction between environmental concerns and monetary policy. Energy consumption is now a direct input into the economic signals that drive rate decisions.
What It Means for Markets
For investors, the BIS paper points to a duality in AI adoption. The technology promises productivity and innovation. Its energy costs could feed broader economic risks if power prices climb on data center demand. If energy costs add to inflationary pressure, central banks could respond with tighter policy — raising rates to cool an economy that is being heated, in part, by the infrastructure required to run AI itself.
The financial stability warning is worth keeping in view. Heavy spending on data centers and new power generation carries the risk of disappointing returns. The BIS has now put that risk on the record. It has not quantified it, and it has not predicted when or how it might materialize. But it has named it.
The paper builds on earlier BIS work published between 2024 and 2026 that analyzed AI’s effects on economic growth, output, and inflation. [1] Paper No. 174 extends that analysis by adding climate to the picture — treating AI’s energy use not as a side issue but as something that feeds directly into the economic signals policymakers watch.
The Consequence Nobody Names
The institutional consequence the BIS paper points toward is not that AI will break the economy. It is that AI may break the instruments we use to understand the economy. Central banks have spent decades building frameworks that assume a certain relationship between energy, output, and prices. AI is changing that relationship in ways that are not yet fully understood.
The paper does not offer a solution. It does not propose new indicators or revised models. It simply describes the problem: the gauges are smudging, and the people who read them need to know.
That is the quiet warning at the center of BIS Paper No. 174. The technology that promises to make so much of the economy smarter is also making it harder to see. And the institutions responsible for steering that economy are only beginning to reckon with what that means.
For central banks, the paper indicates. The environmental effects of AI are not a separate topic to be addressed by climate policy. They are inside the monetary policy framework now, whether policymakers are ready for them or not. The dashboard needs redesigning. The question is whether anyone will do it before the next crisis demands an answer.
For investors, the dual nature of AI adoption — productivity gains paired with energy costs — creates a landscape where the same technology that drives growth can also drive inflation. The BIS has flagged the risk. How quickly the institutions that manage the global economy adapt to a tool that is changing the very signals they rely on remains an open question.
The invisible grid of energy infrastructure that powers AI is now visible in the data. The challenge is learning to read it.