Modern EVs: The untapped powerful computers on wheels

Modern electric vehicles (EVs) have long been described as "batteries on wheels," thanks to their massive energy storage capabilities. However, with advancements in autonomous driving, AI-powered assistance, and high-performance entertainment systems, EVs have also become "GPU/AI-Accelerators on wheels."

These vehicles house unprecedented computational power, yet today, much of it remains locked and therefore unused beyond the purpose of driving. In an era where AI is reshaping industries, society cannot afford to let this potential go to waste.

NVIDIA DRIVE™ hardware [NVIDIA]

The Computational Power Sitting in Cars is Unparalleled

The processing power within EVs is on a scale never seen before in distributed computing. Consider this as an example:

• The NVIDIA DRIVE Orin™ system-on-a-chip (SoC), found in several modern vehicles, delivers up to 254 trillion operations per second (TOPS) [NVIDIA].

• For context, Amazon Web Services (AWS) is developing Ultracluster, a supercomputer powered by its Trainium AI chips, aiming to be one of the most powerful AI training infrastructures [WSJ].

• AWS’s total AI computing power may be estimated (probably badly, but okay within orders of magnitude) at 40–50 ExaOPS, spread across 300–400 data centers and possibly 2–5 million servers worldwide (exact figures are not publicly available).

With this in mind, a simple back-of-the-envelope calculation reveals a startling insight:

If just 150,000 to 200,000 EVs were equipped with an equivalent hardware, their combined computational potential could match AWS’s total AI processing capacity.

That’s less than 0.1% of all passenger cars in Europe—a continent that currently has over 250 million registered vehicles [EUROSTAT].

AI’s Growing Demands Will Surpass Traditional Data Centers

The computational requirements for AI are increasing at an exponential rate. As AI models grow in complexity, centralized cloud providers alone will likely not be able to meet future demand.

Building new data centers is expensive, energy-intensive, and geographically constrained. Meanwhile, EVs already exist, are widely distributed, and sit idle for hours every day—an opportunity waiting to be unlocked.

A Distributed AI Backbone for a Smarter Future

By aggregating the computing power of millions of EVs, it is theoretically possible to create a decentralized, resilient, and energy-efficient AI infrastructure—one that is owned and controlled by people, rather than monopolized by a few corporations.

This shift would not only help meet AI’s growing computational needs, but also reduce reliance on costly, centralized data centers while leveraging surplus clean energy from EV batteries.

It’s Time to Rethink AI Infrastructure

The potential is clear: EVs are more than just vehicles—they are posed to serve as the missing link in AI’s computational revolution. By harnessing the immense power sitting idle in modern cars, one can drive the future of AI in a more sustainable, distributed, and democratized way.

The question is no longer whether we should use EVs as AI accelerators—but when we will seize this opportunity, right now is in the hands of automotive OEMs.