AI Agents Will Drive the Next Revolution

While current AI models are becoming "scary good," the real transformation in the next year will be in the development of agents built on top of these models and the frameworks enabling them, which will be transformative for productivity and learning.

This, and the crypto rails that fuel the agent 2 agent economy, are where you'll find my bets.

Excellent 'cast from a16z dropped yesterday

• The current AI infrastructure build-out is on a scale far exceeding the internet boom of the late 90s and early 2000s, with major geopolitical, economic, and national security implications.
• Demand for AI infrastructure outstrips current supply, creating limitations in power, land, permitting, and supply chains, highlighting the need for far more capacity than projected.
• Data centers are increasingly built where power exists, not vice versa, driving geographic dispersion and greater need for scale-up and scale-out networking, including long-distance interconnects.
• Computing is moving toward extreme hardware specialization, with processors like TPUs offering major efficiency gains over general-purpose CPUs, though architecture cycle times must shrink.
• Geopolitical factors will shape design: China may optimize existing chip tech with abundant power and engineers, while others prioritize efficiency under resource constraints.
• Networking is a growing bottleneck for AI, requiring transformation to manage massive bandwidth demands for both scale-up and scale-out setups, with specialized architectures emerging to match workloads.
• The industry is shifting to deeper co-design partnerships between hardware and software providers, emphasizing open ecosystems over walled gardens to accelerate innovation.
• While current AI models are “scary good,” the next wave of transformation will come from agents and the frameworks enabling them, reshaping productivity and learning.
• Startups should avoid thin wrappers around foundation models and instead integrate them deeply into products, using feedback loops for continual improvement.
• In large organizations, AI tools show big gains in code migration and debugging, though older infrastructure code remains a challenge; a cultural shift toward rapid adoption is essential.
• AI’s evolution drives a “never-ending loop” of demand for more performance and efficiency, as users push for higher quality and longer autonomous execution, fueling continuous innovation.

Let's Discuss! X | LinkedIn