Coinbase CEO Brian Armstrong believes the artificial intelligence industry is heading toward a future where most workloads run on significantly cheaper models, a shift he says is already influencing how crypto firms manage AI costs.
Armstrong shared his outlook during a discussion on rising AI infrastructure expenses, arguing that demand for AI services remains virtually unlimited. However, he expects the market to evolve rapidly, with around 80% of AI workloads moving to models that are up to 99% cheaper than today’s leading systems over the next 12 to 18 months.
Good take
My guess is
– demand for intelligence is near infinite
– but 80% of workloads will be running on 99% cheaper models within 12-18 months
– 20% of workloads will still run on latest gen models where IQ maxing is important (scientific breakthroughs, higher level… https://t.co/gAFtYjorRh— Brian Armstrong (@brian_armstrong) June 8, 2026
According to Armstrong, only a small portion of workloads will continue relying on cutting-edge models. Those advanced systems will remain essential for scientific research, complex reasoning, and high-level AI agents where maximum performance outweighs cost considerations.
Coinbase adopts lower-cost AI strategy
Armstrong revealed that Coinbase is already adjusting its AI operations to take advantage of lower-cost alternatives. The crypto exchange routes many AI requests to more affordable models whenever premium systems are not required.
This approach has helped Coinbase keep AI spending relatively stable despite a sharp increase in usage. Armstrong said the company reserves the most advanced models for specialized tasks while assigning routine operations to cheaper options.
He compared the trend to the personal computer market, where most consumers choose devices that offer sufficient performance at a reasonable price rather than purchasing the most powerful machines available.
Open-source models increase competitive pressure
The discussion gained attention after Delphi Digital co-founder Tommy Shaughnessy highlighted what he sees as a growing challenge for AI companies.
Shaughnessy argued that many AI subscription plans are heavily subsidized and may become difficult to sustain as usage rises. He noted that businesses increasingly have access to open-source models capable of delivering strong performance at a fraction of the cost.
As an example, he pointed to DeepSeek V4, which he said approaches the performance of premium AI systems on software engineering tasks while costing significantly less. Several other open-source models are also narrowing the performance gap with leading proprietary systems.
AI economics could reshape the industry
Providers such as OpenRouter, Venice, Baseten, and Together has expanded access to lower-cost AI models while supporting enterprise privacy and compliance requirements.
Armstrong believes the dropping inference costs and continued improvements in open-source AI will encourage more businesses to adopt a hybrid approach. Rather than paying premium prices for every task, companies can deploy advanced models only where necessary and rely on cheaper alternatives for everyday operations.
Meanwhile, Armstrong laid out what he believes are the biggest upgrades still needed in the global financial system, arguing that technology and policy must evolve together to unlock the next stage of growth.
Enjoyed this? Bookmark DeFi Planet, explore related topics, and follow us on Twitter, LinkedIn, Facebook, Instagram, Threads and CoinMarketCap Community for seamless access to high-quality industry insights
Take control of your crypto portfolio with DEFI PLANET PRO, DeFi Planet’s suite of analytics tools
