OpenAI Just Forecast a $280 Billion Burn. Marketing Handed the Buy to the Algorithm.

OpenAI told investors it could burn nearly $280 billion by 2030.

Same week, the IAB raised the US ad-spend forecast because AI is now buying the ads.

That is not two stories. That is one market. The labs are spending like they intend to own the channel. Marketing is handing the buy to the algorithm. Agencies still treating AI as a cheaper caption machine are on the wrong side of both.

What actually shipped this week

The Financial Times reported OpenAI expects to burn about $278 billion in cash between 2026 and 2030 as it pays for compute and infrastructure. The same week it was talking with investors about a valuation near $1.2 trillion ahead of a possible listing.

That number is not a flex. It is a statement of intent. You do not budget that kind of burn to stay a writing assistant.

On the marketing side, the Interactive Advertising Bureau revised 2026 US ad-spend growth up from 9.5 percent to 12.3 percent. The driver was not a new format. It was automated, on-platform AI buying. Meta Advantage+. Google Performance Max. The same class of tools on TikTok, Reddit, and Pinterest.

Media leads talking to Digiday put 11 to 12 percent of digital spend already inside those tools. Some books are at 30 percent. Every major ad platform now ships an MCP server, so Claude, Gemini, or ChatGPT can talk to the account in plain language and move budget.

Marketing and AI: the cheap-content trap

Here is the part most recaps skip, and it is the one that hits agency P&L first.

Forrester already found nine in ten US marketing agencies use generative AI. Half use agentic AI for execution. The main reason is still staff productivity and cost. Not better work. Faster, cheaper work.

That is a bad trade in a week when the platforms are taking the media plan and the labs are pricing themselves like utilities. If your AI stack is forty captions before lunch, you are funding someone else’s channel.

We are running it differently at Dzine Prodigy.

LAYER 1 — Own the offer the algorithm can spend against. Price, proof, next step, do-not-say list. If Performance Max cannot see a clean offer, it will invent one.

LAYER 2 — Volume with a human taste pass. Generate the cuts. Kill the ones that sound like every other brand in the auction.

LAYER 3 — Treat AI buying platforms as a channel with a brief, not a black box. Someone on the team owns the objective the model is actually optimizing.

Do this week

Pick one live campaign. Write the 10-line brief the algorithm is optimizing against: audience, offer, proof, exclusions, the one action that counts as a win. If you cannot write it, the tool is running the account.

The labs will keep burning cash to sit between you and the buyer. Finish the brief before they finish the channel.