OpenAI Tests Ads on ChatGPT, Acquires Health Startup Torch

Two significant moves from OpenAI this week signal how the company is thinking about revenue diversification as operational costs continue to climb: a coming ad rollout on ChatGPT, and a $100 million acquisition in healthcare.

Key points

• OpenAI will test advertisements on ChatGPT targeting free and lower-tier subscribers in the US.
• The company says ads will not influence ChatGPT’s responses or compromise user privacy.
• OpenAI acquired healthcare startup Torch for $100 million to support its ChatGPT Health product.
• Torch’s technology reportedly helps unify patient records through a secure AI “medical memory” layer.
• Both moves reflect pressure on OpenAI to build revenue streams beyond its core subscription tiers, where paying subscriber counts remain limited.

Why ads, and why now?

OpenAI’s costs are substantial. Running large-scale inference on models like GPT-4o and its successors requires enormous compute spend, and subscription revenue from ChatGPT’s free and lower-paying tiers does not cover that load. The ad test, starting in the US, is a direct attempt to monetize that large base of non-paying or lightly paying users.

The company has been explicit that answers will remain unbiased and that user privacy will not be affected. Those assurances are easy to make at launch and harder to verify over time, which is why investor observers will want to watch how the ad product is structured. The central tension is real: advertisers pay to reach audiences, and the more useful that targeting is, the more user data has to factor in somewhere.

The competitive backdrop matters here too. Google and Meta have mature, sophisticated ad businesses built on years of behavioral data and measurement infrastructure. OpenAI is starting from scratch in a format, conversational AI, that does not yet have established norms for ad placement or measurement. Whether ChatGPT can generate meaningful ad revenue quickly enough to matter on the balance sheet is an open question. The bet is that a large, engaged user base interacting with a trusted assistant creates ad inventory worth paying for.

What does the Torch acquisition add to ChatGPT Health?

OpenAI launched ChatGPT Health earlier this month (covered in our January 11 edition). The Torch acquisition, priced at $100 million, is the follow-on infrastructure play. Torch’s core capability appears to be record unification: pulling together fragmented patient data into a coherent picture that an AI system can reason over securely.

For ChatGPT Health to be genuinely useful in clinical or consumer health contexts, it needs persistent, accurate context about a user’s medical history. A system that forgets what it knew last session, or that cannot connect data across providers, has limited practical value. Torch’s “medical memory” framing suggests OpenAI is trying to solve exactly that continuity problem.

Healthcare is also one of the few sectors where users have demonstrated willingness to pay a meaningful premium for software that works reliably. If OpenAI can establish trust in that space, it opens a higher-value subscription tier than the general ChatGPT consumer plans. At $100 million, the Torch deal is not small, but it is modest relative to OpenAI’s overall funding scale, and it buys the company a team and technology it would have taken considerably longer to build internally.

What this means for the IPO picture

OpenAI has not set a public timeline for an IPO, but the structural logic of both moves points toward preparing a business that looks credible to public market investors. An ad business adds a revenue line that scales with usage rather than requiring users to convert to paid plans. A healthcare vertical adds a defensible, high-margin opportunity in a large market.

The risk is execution. Advertising requires sales relationships, measurement tools, and policies around content adjacency that OpenAI does not yet have at scale. Healthcare requires regulatory navigation, clinician trust, and security infrastructure that takes years to build properly. Both are the right strategic directions; neither is straightforward to pull off. Investors tracking OpenAI’s path to a public offering should treat this week’s announcements as directional signals, not confirmed revenue inflection points.