OpenAI to Charge $60 CPM on Ads with a $200k Minimum Commitment

Key Takeaways

• $60 CPM far exceeds typical digital ad rates
• Minimum spend set at $200,000 for beta participants
• Advertisers receive only impressions and clicks metrics
• Some brands offered lower commitments around $100k‑$125k
• OpenAI’s ad rollout appears rushed for enterprise clients

Pulse Analysis

OpenAI’s decision to launch a paid advertising slot inside ChatGPT marks a bold pivot from its traditional AI‑as‑a‑service model. By pricing the inventory at roughly $60 per thousand views, the company is positioning its ad inventory alongside premium live‑sports broadcasts, a rate that dwarfs the roughly $20 CPM that dominates most social and search platforms. The premium reflects both the perceived value of a conversational interface and OpenAI’s ambition to create a new revenue stream, but it also sets a high bar for advertisers accustomed to more cost‑effective digital channels.

The beta program’s entry barrier is equally steep, with a $200,000 minimum spend—though some reports indicate negotiations as low as $100,000 to $125,000. In exchange, advertisers receive only surface‑level metrics such as total impressions and click counts, lacking the granular attribution data that modern marketers rely on to tie spend to conversions. This limited visibility makes it difficult to measure true performance, raising doubts about ROI and potentially slowing adoption among risk‑averse enterprises that already allocate sizable budgets to proven platforms like Google and Meta.

If OpenAI can leverage its AI expertise to deliver richer, real‑time attribution, the platform could eventually justify its premium pricing and reshape how brands reach consumers in conversational spaces. Competitors may follow suit, prompting a wave of AI‑enhanced ad solutions that blend contextual relevance with measurable outcomes. Until then, the current rollout feels more like an experimental test than a fully‑fledged offering, and the market will be watching closely to see whether the high‑cost, low‑data model can gain traction or be forced to evolve.