Artificial intelligence is changing the world—but behind the innovation is a not-so-small energy bill.
To understand why AI models, especially large language models (LLMs) like GPT‑3.5, require so much power, we turned to David Kerr, Forcura’s former Chief Technology Officer, for a quick breakdown.
“When you train a large language model, you have to use thousands of different types of chips—GPUs or TPUs—and they consume a tremendous amount of energy. It's estimated that training GPT‑3.5 used enough electricity to power a single-family home for 120 years,” says David.
So...Why All the Watts?
Training a generative AI model isn’t as simple as pressing “start.” It involves massive computing infrastructure— thousands of GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units) working overtime, often for weeks on end. These chips are built for speed and smarts, crunching data so AI can learn to understand language, spot patterns, and make decisions.
But here’s the catch: all that brainpower generates a lot of heat. And just like your laptop gets hot after a long Teams call, these systems need constant cooling. Data centers must maintain optimal conditions to prevent overheating, adding a second layer of energy demand. So it’s not just about powering the brains—it’s also about keeping them cool.
What It Means for AI in Healthcare
As healthcare adopts more AI-powered tools, energy usage becomes part of the conversation. At Forcura, we’re committed to responsible AI in healthcare—developing smarter, more efficient solutions that support providers and sustainability.
We do this by:
Using smaller, focused models trained on relevant healthcare data
Applying transfer learning to reduce compute time
Partnering with cloud providers that prioritize renewable energy
This way, we can deliver powerful tools like Forcura IQ and Referral Summary while keeping energy use in check.
🔍 This post is part of our AI in Healthcare Blog Series.
Want to go deeper into topics like generative AI, large language models, and their impact on healthcare? Visit our AI Authority page.
🎥 Watch David explain why AI uses so much energy in this short video.