INFER/CON
The Inference conference. Fall 2026.
When inference becomes production infrastructure.
MONDAY, NOVEMBER 16, 2026
CONVENE · SAN FRANCISCO, CA
********
Invite-Only and Application-Based. Seats are limited by design.
A curated practitioner audience - builders first.
Enter your email to apply. If accepted, you'll be notified when session registration begins.
********
INFER/CON is an industry conference in San Francisco focused on the operational realities of running AI at scale.
As AI moves from experimentation to core production infrastructure, companies still solve pieces of the problem in isolation. INFER/CON brings them into one room to explore:
How do we run AI systems reliably, efficiently, and securely? What happens when inference becomes production infrastructure at global scale?
********
Who is the audience?
Built for and by the people building and operating the AI stack — platform and infrastructure engineers, architects, technical founders, model and inference providers, compute and chip companies, and the teams running the largest production AI workloads today.
********
The Conversation
AI Reliability Engineering · Agent Observability · Inference Economics · Latency Optimization · Evaluation & Monitoring · AI Security & Governance · Multi-model, Multi-provider Systems at Scale
Full agenda to be disclosed soon.
********
The Line-Up
Leadership and founders from the top companies contributing to the inference across the stack.
Interested in sponsoring InferCon 2026? Contact Us
********
How is it different from other AI conferences?
Most AI conferences are about building models or applications. INFER/CON is about operating them, and what happens after the model ships: reliability, cost, latency, security, and scale.
********
Why does this event need to exist?
AI is becoming a foundational infrastructure layer. The industry has plenty of opportunities to share research or specific company products, and needs more forums to share best practices around operating inference in production at scale. Vendors solve pieces of the problem in isolation. INFER/CON provides a neutral venue where the people building the AI stack can combine and compare notes on what breaks in production and what works.