The Hidden Waste Behind AI Compute with Justin Gelinas
July 25, 2026
Damien Filiatrault talks with Justin Gelinas, Chief Business Officer at TAHO Labs, about the hidden inefficiencies driving up the cost and energy demands of AI infrastructure. They explore how better workload distribution, improved hardware utilization, and more efficient compute could help companies get more from existing resources while unlocking new opportunities for AI and software development.
Host Damien Filiatrault talks with Justin Gelinas, Chief Business Officer at TAHO Labs, about the hidden inefficiencies behind AI infrastructure and why the industry may need to get more from the hardware it already has.
Justin explains how TAHO is building low-level software that breaks large workloads into smaller jobs and distributes them across available processors. The goal is to keep GPUs and other resources working more effectively, reduce idle capacity, and support different types of hardware within the same environment.
The conversation also explores Wirth’s law and Jevons paradox: why faster hardware does not always produce faster software, and why lowering the cost of compute may increase overall demand rather than reduce it. Damien and Justin connect these ideas to AI development, infrastructure costs, energy constraints, and the future of software engineering.
Justin also shares why TAHO uses AI tools extensively for documentation and operational work, but less for its core code. When engineers are inventing techniques without established examples, coding agents can quickly reach their limits or introduce unnecessary technical debt.
For developers, technology leaders, and anyone interested in the economics of AI, this episode offers a closer look at what is happening beneath the application layer—and why efficiency may become one of the industry’s most important challenges.
What you’ll learn
- What Wirth’s law and Jevons paradox reveal about software and compute
- Why cheaper, more efficient AI may create more demand
- Where AI coding tools struggle with genuinely new technical problems
- How TAHO divides and distributes large workloads
- Why interoperability across different hardware providers matters
- Why many compute environments use only a fraction of their available capacity
- How energy and infrastructure constraints are affecting AI growth
- Why improving compute efficiency can reduce costs and create room for innovation
Memorable sound bites
“Software gets slower more rapidly than hardware becomes faster.”
“When we can produce more value at a fraction of the cost, that creates opportunities.”
“When you are inventing a genuinely new technique, AI can reach a brick wall quickly.”
“Those constraints make efficiency unavoidable.”
“Innovation and efficiency can reinforce each other.”
Links
Get 20% off your first month with Scalable Path: https://www.scalablepath.com/commit
Commit & Push Website: https://www.commit-push.com/
Scalable Path Website: https://www.scalablepath.com/