ChatGPT's advertising business is expanding quickly, but a growing advertiser base has yet to deliver the revenue OpenAI expected. Dan Salmon, a partner at New Street Research, estimates that advertiser numbers climbed from around 20,000 at the end of August 2026 to roughly 60,000 in early October. Yet OpenAI is likely to miss the $2.4 billion advertising revenue forecast it presented to investors for 2026, according to The Information.

Lower entry costs are helping businesses experiment. A comparison reported by MarTech puts ChatGPT clicks at about $5, versus 30–50 for similar commercial software ads on Google. OpenAI now needs to turn that willingness to test into recurring advertising expenditure.

Easier entry and a wider geographic reach

ChatGPT advertising began as a closed pilot for a limited group of large brands. Advertisers paid per thousand impressions and faced a minimum commitment of $200,000. Even a trial campaign therefore required a substantial budget decision.

The platform subsequently introduced payment per click. It now supports CPM and CPC bidding, with recommended CPC bids of 3–5. Industry reports also describe tests of cost-per-action bidding with selected advertisers, although OpenAI has not publicly confirmed that pricing model.

Self-service purchasing through Ads Manager is expanding too. In September 2026, ads reached Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan. Businesses in those markets could use the buying interface from launch, taking the service's overall availability beyond 60 countries.

The expansion also calls for care when interpreting OpenAI's statement that advertising reached a $1 billion annualized revenue run rate less than 200 days after launch. That figure projects a current level of revenue across 12 months; it does not represent revenue actually earned over a year. MarTech notes that the company provided no independent verification of the figure. Both OpenAI and Anthropic use such annualized calculations.

Conversation as the basis for ad targeting

When selecting an ad, ChatGPT can consider the context and intent of the current conversation, advertiser descriptions of relevant topics, the ad itself and its landing page. Additional signals may be used when personalization is enabled.

Rather than relying solely on exact keyword matches, businesses can supply context hints describing topics and situations in which their offer would be relevant. The approach combines signals of immediate user intent with audience tools familiar to advertisers.

Businesses can also upload email addresses and phone numbers to match customer lists. According to Gadget Review's account of OpenAI's documentation, audiences used for this targeting must contain at least 25,000 matched users.

The surrounding context can change with every message, however. OpenAI is working with DoubleVerify and Integral Ad Science to evaluate placement suitability and says it has safeguards against ads appearing in sensitive conversations or situations of emotional vulnerability. The company says those partners do not receive access to private conversations.

From an ad to a conversation with a brand

In the United States, OpenAI is testing Sponsored Agents, which let users click an ad and enter a separate conversation with an advertiser's AI agent. Someone considering a dining table could ask whether it fits their room, follow up with further questions and then visit the seller's website.

This adds a step between initial interest and the website visit. Marketing teams must decide what information the agent holds, which claims it may make, how product details stay current and when the conversation should be handed off. Measuring the interaction becomes another responsibility.

Ads Manager has also gained HubSpot and Shopify integrations, alongside tools for creating and editing campaigns through natural-language requests. An optional feature can adapt existing headlines and descriptions to the conversation. These capabilities connect ChatGPT advertising more closely with other marketing systems while adding complexity to campaign oversight and messaging controls.

Visual formats and the test of effectiveness

In October, OpenAI announced a US test of visual ads alongside image-generation results. It says the placements will be clearly labeled, kept separate from the image being created and have no influence on ChatGPT's answers. Such ads could fit situations in which someone is developing an idea for a product, trip, event or design.

Measurement capabilities are developing in parallel. Conversion-data partners include Hightouch, Tealium and LiveRamp, while attribution and measurement providers include AppsFlyer, Adjust, Branch, Triple Whale, Kochava, Haus, Measured and others.

Those tools should help establish whether advertiser growth reflects repeat buying or a large number of inexpensive trials. The decision to allocate an ongoing budget will depend on conversions, the additional outcomes advertising produces and the quality of acquired customers compared with other channels.

ChatGPT's advertising business is expanding quickly, but a growing advertiser base has yet to deliver the revenue OpenAI expected. Dan Salmon, a partner at New Street Research, estimates that advertiser numbers climbed from around 20,000 at the end of August 2026 to roughly 60,000 in early October. Yet OpenAI is likely to miss the $2.4 billion advertising revenue forecast it presented to investors for 2026, according to The Information.

Lower entry costs are helping businesses experiment. A comparison reported by MarTech puts ChatGPT clicks at about $5, versus 30–50 for similar commercial software ads on Google. OpenAI now needs to turn that willingness to test into recurring advertising expenditure.

Easier entry and a wider geographic reach

ChatGPT advertising began as a closed pilot for a limited group of large brands. Advertisers paid per thousand impressions and faced a minimum commitment of $200,000. Even a trial campaign therefore required a substantial budget decision.

The platform subsequently introduced payment per click. It now supports CPM and CPC bidding, with recommended CPC bids of 3–5. Industry reports also describe tests of cost-per-action bidding with selected advertisers, although OpenAI has not publicly confirmed that pricing model.

Self-service purchasing through Ads Manager is expanding too. In September 2026, ads reached Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan. Businesses in those markets could use the buying interface from launch, taking the service's overall availability beyond 60 countries.

The expansion also calls for care when interpreting OpenAI's statement that advertising reached a $1 billion annualized revenue run rate less than 200 days after launch. That figure projects a current level of revenue across 12 months; it does not represent revenue actually earned over a year. MarTech notes that the company provided no independent verification of the figure. Both OpenAI and Anthropic use such annualized calculations.

Conversation as the basis for ad targeting

When selecting an ad, ChatGPT can consider the context and intent of the current conversation, advertiser descriptions of relevant topics, the ad itself and its landing page. Additional signals may be used when personalization is enabled.

Rather than relying solely on exact keyword matches, businesses can supply context hints describing topics and situations in which their offer would be relevant. The approach combines signals of immediate user intent with audience tools familiar to advertisers.

Businesses can also upload email addresses and phone numbers to match customer lists. According to Gadget Review's account of OpenAI's documentation, audiences used for this targeting must contain at least 25,000 matched users.

The surrounding context can change with every message, however. OpenAI is working with DoubleVerify and Integral Ad Science to evaluate placement suitability and says it has safeguards against ads appearing in sensitive conversations or situations of emotional vulnerability. The company says those partners do not receive access to private conversations.

From an ad to a conversation with a brand

In the United States, OpenAI is testing Sponsored Agents, which let users click an ad and enter a separate conversation with an advertiser's AI agent. Someone considering a dining table could ask whether it fits their room, follow up with further questions and then visit the seller's website.

This adds a step between initial interest and the website visit. Marketing teams must decide what information the agent holds, which claims it may make, how product details stay current and when the conversation should be handed off. Measuring the interaction becomes another responsibility.

Ads Manager has also gained HubSpot and Shopify integrations, alongside tools for creating and editing campaigns through natural-language requests. An optional feature can adapt existing headlines and descriptions to the conversation. These capabilities connect ChatGPT advertising more closely with other marketing systems while adding complexity to campaign oversight and messaging controls.

Visual formats and the test of effectiveness

In October, OpenAI announced a US test of visual ads alongside image-generation results. It says the placements will be clearly labeled, kept separate from the image being created and have no influence on ChatGPT's answers. Such ads could fit situations in which someone is developing an idea for a product, trip, event or design.

Measurement capabilities are developing in parallel. Conversion-data partners include Hightouch, Tealium and LiveRamp, while attribution and measurement providers include AppsFlyer, Adjust, Branch, Triple Whale, Kochava, Haus, Measured and others.

Those tools should help establish whether advertiser growth reflects repeat buying or a large number of inexpensive trials. The decision to allocate an ongoing budget will depend on conversions, the additional outcomes advertising produces and the quality of acquired customers compared with other channels.

ChatGPT's advertising business is expanding quickly, but a growing advertiser base has yet to deliver the revenue OpenAI expected. Dan Salmon, a partner at New Street Research, estimates that advertiser numbers climbed from around 20,000 at the end of August 2026 to roughly 60,000 in early October. Yet OpenAI is likely to miss the $2.4 billion advertising revenue forecast it presented to investors for 2026, according to The Information.

Lower entry costs are helping businesses experiment. A comparison reported by MarTech puts ChatGPT clicks at about $5, versus 30–50 for similar commercial software ads on Google. OpenAI now needs to turn that willingness to test into recurring advertising expenditure.

Easier entry and a wider geographic reach

ChatGPT advertising began as a closed pilot for a limited group of large brands. Advertisers paid per thousand impressions and faced a minimum commitment of $200,000. Even a trial campaign therefore required a substantial budget decision.

The platform subsequently introduced payment per click. It now supports CPM and CPC bidding, with recommended CPC bids of 3–5. Industry reports also describe tests of cost-per-action bidding with selected advertisers, although OpenAI has not publicly confirmed that pricing model.

Self-service purchasing through Ads Manager is expanding too. In September 2026, ads reached Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan. Businesses in those markets could use the buying interface from launch, taking the service's overall availability beyond 60 countries.

The expansion also calls for care when interpreting OpenAI's statement that advertising reached a $1 billion annualized revenue run rate less than 200 days after launch. That figure projects a current level of revenue across 12 months; it does not represent revenue actually earned over a year. MarTech notes that the company provided no independent verification of the figure. Both OpenAI and Anthropic use such annualized calculations.

Conversation as the basis for ad targeting

When selecting an ad, ChatGPT can consider the context and intent of the current conversation, advertiser descriptions of relevant topics, the ad itself and its landing page. Additional signals may be used when personalization is enabled.

Rather than relying solely on exact keyword matches, businesses can supply context hints describing topics and situations in which their offer would be relevant. The approach combines signals of immediate user intent with audience tools familiar to advertisers.

Businesses can also upload email addresses and phone numbers to match customer lists. According to Gadget Review's account of OpenAI's documentation, audiences used for this targeting must contain at least 25,000 matched users.

The surrounding context can change with every message, however. OpenAI is working with DoubleVerify and Integral Ad Science to evaluate placement suitability and says it has safeguards against ads appearing in sensitive conversations or situations of emotional vulnerability. The company says those partners do not receive access to private conversations.

From an ad to a conversation with a brand

In the United States, OpenAI is testing Sponsored Agents, which let users click an ad and enter a separate conversation with an advertiser's AI agent. Someone considering a dining table could ask whether it fits their room, follow up with further questions and then visit the seller's website.

This adds a step between initial interest and the website visit. Marketing teams must decide what information the agent holds, which claims it may make, how product details stay current and when the conversation should be handed off. Measuring the interaction becomes another responsibility.

Ads Manager has also gained HubSpot and Shopify integrations, alongside tools for creating and editing campaigns through natural-language requests. An optional feature can adapt existing headlines and descriptions to the conversation. These capabilities connect ChatGPT advertising more closely with other marketing systems while adding complexity to campaign oversight and messaging controls.

Visual formats and the test of effectiveness

In October, OpenAI announced a US test of visual ads alongside image-generation results. It says the placements will be clearly labeled, kept separate from the image being created and have no influence on ChatGPT's answers. Such ads could fit situations in which someone is developing an idea for a product, trip, event or design.

Measurement capabilities are developing in parallel. Conversion-data partners include Hightouch, Tealium and LiveRamp, while attribution and measurement providers include AppsFlyer, Adjust, Branch, Triple Whale, Kochava, Haus, Measured and others.

Those tools should help establish whether advertiser growth reflects repeat buying or a large number of inexpensive trials. The decision to allocate an ongoing budget will depend on conversions, the additional outcomes advertising produces and the quality of acquired customers compared with other channels.

ChatGPT's advertising business is expanding quickly, but a growing advertiser base has yet to deliver the revenue OpenAI expected. Dan Salmon, a partner at New Street Research, estimates that advertiser numbers climbed from around 20,000 at the end of August 2026 to roughly 60,000 in early October. Yet OpenAI is likely to miss the $2.4 billion advertising revenue forecast it presented to investors for 2026, according to The Information.

Lower entry costs are helping businesses experiment. A comparison reported by MarTech puts ChatGPT clicks at about $5, versus 30–50 for similar commercial software ads on Google. OpenAI now needs to turn that willingness to test into recurring advertising expenditure.

Easier entry and a wider geographic reach

ChatGPT advertising began as a closed pilot for a limited group of large brands. Advertisers paid per thousand impressions and faced a minimum commitment of $200,000. Even a trial campaign therefore required a substantial budget decision.

The platform subsequently introduced payment per click. It now supports CPM and CPC bidding, with recommended CPC bids of 3–5. Industry reports also describe tests of cost-per-action bidding with selected advertisers, although OpenAI has not publicly confirmed that pricing model.

Self-service purchasing through Ads Manager is expanding too. In September 2026, ads reached Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan. Businesses in those markets could use the buying interface from launch, taking the service's overall availability beyond 60 countries.

The expansion also calls for care when interpreting OpenAI's statement that advertising reached a $1 billion annualized revenue run rate less than 200 days after launch. That figure projects a current level of revenue across 12 months; it does not represent revenue actually earned over a year. MarTech notes that the company provided no independent verification of the figure. Both OpenAI and Anthropic use such annualized calculations.

Conversation as the basis for ad targeting

When selecting an ad, ChatGPT can consider the context and intent of the current conversation, advertiser descriptions of relevant topics, the ad itself and its landing page. Additional signals may be used when personalization is enabled.

Rather than relying solely on exact keyword matches, businesses can supply context hints describing topics and situations in which their offer would be relevant. The approach combines signals of immediate user intent with audience tools familiar to advertisers.

Businesses can also upload email addresses and phone numbers to match customer lists. According to Gadget Review's account of OpenAI's documentation, audiences used for this targeting must contain at least 25,000 matched users.

The surrounding context can change with every message, however. OpenAI is working with DoubleVerify and Integral Ad Science to evaluate placement suitability and says it has safeguards against ads appearing in sensitive conversations or situations of emotional vulnerability. The company says those partners do not receive access to private conversations.

From an ad to a conversation with a brand

In the United States, OpenAI is testing Sponsored Agents, which let users click an ad and enter a separate conversation with an advertiser's AI agent. Someone considering a dining table could ask whether it fits their room, follow up with further questions and then visit the seller's website.

This adds a step between initial interest and the website visit. Marketing teams must decide what information the agent holds, which claims it may make, how product details stay current and when the conversation should be handed off. Measuring the interaction becomes another responsibility.

Ads Manager has also gained HubSpot and Shopify integrations, alongside tools for creating and editing campaigns through natural-language requests. An optional feature can adapt existing headlines and descriptions to the conversation. These capabilities connect ChatGPT advertising more closely with other marketing systems while adding complexity to campaign oversight and messaging controls.

Visual formats and the test of effectiveness

In October, OpenAI announced a US test of visual ads alongside image-generation results. It says the placements will be clearly labeled, kept separate from the image being created and have no influence on ChatGPT's answers. Such ads could fit situations in which someone is developing an idea for a product, trip, event or design.

Measurement capabilities are developing in parallel. Conversion-data partners include Hightouch, Tealium and LiveRamp, while attribution and measurement providers include AppsFlyer, Adjust, Branch, Triple Whale, Kochava, Haus, Measured and others.

Those tools should help establish whether advertiser growth reflects repeat buying or a large number of inexpensive trials. The decision to allocate an ongoing budget will depend on conversions, the additional outcomes advertising produces and the quality of acquired customers compared with other channels.