Why it costs AI much more to deliver an ad than Facebook or Google

The cost gap does not come from ads. It comes from content that creates the environment for the ads.

In search, social media and AI systems, ads are prebuilt assets selected and delivered. The difference is who creates the content environment those ads appear in — and at what cost.

Social and search platforms get their content environment for free

The experience exists because others create the content.

On platforms like Facebook, YouTube, and Google, the platform does not create the underlying content.

• In social products, users create posts, videos, comments, and images.
• In search, independent publishers create webpages, articles, and media that Google merely indexes and ranks.

In both cases:

• The content arrives fully formed.
• The platform does not pay to create it.
• It is stored once and delivered many times.

The most expensive part of the experience — content creation — is outsourced to users and publishers for free. So search and social differ in interface, but their economics are identical: both monetize access to content they did not need to pay to produce.

But AI platforms must create the content environment themselves. There is no free content supply.

• Systems like ChatGPT do not receive user-generated answers.
• The “feed” is the response itself.
• Every interaction requires the system to generate new content on demand.

Even if generation were cheap, it is still a cost social platforms do not have.

But content creation in AI is not cheap — it is extremely expensive

This is where the economics break.

• Generating language requires large-scale inference on GPUs.
• Each response involves billions to trillions of operations.
• Power, hardware depreciation, memory bandwidth, and latency all apply.

This cost is many orders of magnitude greater than simply delivering existing content

Delivery costs are trivial in all cases

• Once content exists, delivery over a CDN is cheap.
• That is true for social posts, ads, and AI responses alike.
• Delivery is not the problem.

The problem is the cost of creating what gets delivered.

Social ads ride on free content. AI ads ride on paid content.
This is the structural asymmetry.

• Social ads are inserted into a stream that already exists.
• AI ads must be inserted into a response that had to be computed.
• The ad itself is cheap. The surrounding content is not.

Advertising does not pay for the thing that costs the most.

Scale helps social platforms. It hurts AI platforms.
The economics move in opposite directions.

• More social users means more free content and lower average cost.
• More AI users means more generated content and higher total compute spend.
• There is no free supply curve to exploit.

Growth amplifies cost instead of diluting it.

This is not a temporary inefficiency. It is structural.
No optimization changes the core fact.

• Social platforms do not pay to create their content environment.
• AI platforms must.
• And the cost of creating that environment dwarfs delivery.

As long as AI systems generate content on demand, this cost remains.

The bottom line.
Ads are prebuilt assets in both social and AI systems.
The difference is that social platforms get their content for free.
AI platforms must compute it — and that computation costs orders of magnitude more than delivering it.

That is why AI ads are fundamentally more expensive than social ads — and why AI economics cannot look like social media economics.