Breaking

AT&T, Airbnb and Deloitte Signal a Corporate Shift Toward Open AI Models

September 4, 2026 at 12:37 EDT

Foundation Models
Sales & Business
Open Source

U.S. companies are increasingly mixing open and proprietary AI models as token costs rise and buyers seek greater control over deployment and data. AT&T is routing production workloads to cheaper models, Airbnb is using open-weight technology in customer support, and Deloitte has launched a practice to help clients build hybrid AI systems.

SEP 2026 · ENTERPRISE AI

Open AI Models Move Into the Enterprise Mix

AT&T says its gateway can cut AI costs by up to 90%; Airbnb and Deloitte show how open models are moving into enterprise architecture.

The cost-and-control pivot

45B

AT&T tokens per day

90%

maximum cost reduction

~1/3

Airbnb North American support issues handled by AI

80%

Airbnb engineers using AI tools

AT&T’s maximum stated reduction

Cache-aware routing sends each task to a model matched to its cost, speed and expected quality.

90%

stated maximum reduction

Nine of ten segments mark the stated maximum

The visual uses ten equal blocks so the reported percentage is readable at a glance.

Why teams adopt open models

Lower running cost at scale
More control over data and deployment
Fine-tuning for specialized tasks

Where friction remains

Infrastructure and operations still cost money
Training data may remain opaque
Security and provenance concerns continue

THE ENTERPRISE PLAYBOOK

The practical architecture is mixed-model

Keep frontier models for demanding workloads. Route routine, high-volume or specialized tasks to open models when cost and control matter.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.