Twenty-five percent. That is how much Google’s total carbon emissions have climbed in just one year. Amazon’s footprint is up 16 percent. For two companies that have built their public identities around aggressive net-zero pledges, these numbers are not just a setback. They are a warning.

For years, Big Tech’s climate strategy was simple: build data centers, buy massive amounts of renewable energy, and call it a day. It worked. But the AI boom has upended that math. The energy required to train and run large language models is staggering. It is not just the electricity powering the servers; it is the physical cost of the infrastructure itself. The dream of a carbon-neutral future is colliding with the reality of a massive, power-hungry hardware build-out.

The Scope 3 Problem

Most of the damage is hidden in what companies call Scope 3 emissions. This category covers the pollution a company does not directly control, such as the manufacturing of the goods it buys or the products it sells. For Google and Amazon, this includes the thousands of GPUs and memory chips required to keep AI models humming.

Semiconductor manufacturing is a dirty business. Many of the world’s leading-edge chip factories are located in regions where electrical grids remain heavily dependent on fossil fuels. Worse, the chemical processes used to etch these chips involve gases that are thousands of times more potent than CO2. When Amazon adds 1.2 gigawatts of data center capacity in a single quarter, it is not just buying servers. It is buying the carbon debt of every factory that produced those components.

Why Renewables Aren't Enough

For a decade, tech giants relied on wind and solar to offset their growth. That strategy is failing. AI demands constant, reliable power—the kind that intermittent renewables struggle to provide on their own. As a result, companies are increasingly turning to natural gas to keep their data centers from flickering out.

Then there is the construction. Data centers are essentially fortresses of steel and concrete. Both materials are notoriously carbon-intensive to produce. While startups are experimenting with greener alternatives, they cannot yet deliver at the scale required by a company like Amazon. The sheer volume of construction needed to house the next generation of AI hardware is creating a carbon floor that is becoming impossible to lower.

What This Means for the Future

Neither company explicitly blames AI for the surge in their reports, but the correlation is undeniable. They are now in a race against their own growth. To hit their net-zero targets, they will need to do more than just buy carbon credits. They will need to fundamentally change how they build.

This will require heavy investment in advanced manufacturing for steel and cement. It will mean pushing for cleaner grids in the regions where their chips are made. It might even mean slowing down the pace of expansion. The current trajectory is unsustainable. If they cannot decouple AI growth from carbon emissions, their climate pledges will become nothing more than marketing collateral.

Key Takeaways

  • Emissions are surging: Google and Amazon reported double-digit increases in carbon emissions, driven largely by the infrastructure needs of AI.
  • The Scope 3 trap: Most of the new pollution comes from the manufacturing of hardware and data center construction, rather than direct energy use.
  • The fossil fuel fallback: Because AI requires constant power, tech giants are increasingly relying on natural gas rather than relying solely on renewables.

What happens next depends on the balance sheet. If the cost of decarbonizing their supply chain exceeds the value of their AI services, the climate goals will likely be the first thing to go. The next round of sustainability reports will show if these companies are serious about their promises or if they are simply waiting for the public to stop looking.