OpenAI crossed $1 billion in annualized ad revenue in under 200 days, and this week the platform went live in Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam, and Taiwan. That is every core Southeast Asian market in one announcement, and it pushes ChatGPT Ads past 60 countries total.
If you run marketing in this region, nobody outside OpenAI has enough regional data to tell you whether the channel works. The useful question is earlier than that: do you build the muscle now, while the auction is thin and mistakes are cheap, or wait two quarters and pay retail for the same education?
What actually shipped
Seven new markets went live at once: Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam, and Taiwan. The launch follows earlier Asia-Pacific rollouts in Australia, New Zealand, Japan, Korea, and India, so this was the logical next block rather than a surprise. Still, the scale of the single drop matters. One week you had no access; now your entire regional footprint can be covered from one Ads Manager account.
Two details in the announcement deserve more attention than the country list. Shopee is named as the strategic launch partner for the region, extending a collaboration that started in Brazil. And tens of thousands of advertisers are already running campaigns globally. Regional e-commerce is not watching this from the sidelines. Shopee treating ChatGPT as a performance channel, not a science experiment, is the strongest signal in the whole release.
Who sees the ads, and who doesn’t
Ads appear only for users on the Free and Go plans. Plus, Pro, and Enterprise accounts stay ad-free, and OpenAI repeats the same principles it launched with: conversations stay private from advertisers, ads are labeled and kept separate from answers, and personalization is user-controlled.
For advertisers, this cuts two ways. The Free and Go audience is huge and includes exactly the price-sensitive shoppers most SEA commerce brands want. But it also means the ad sits inside a conversation someone is actively having with a tool they trust. Intent is high; patience is low. Creative that interrupts will burn the placement. The formats that survive here will look less like banners and more like useful answers to the question the person just asked.
Reading the $1 billion properly
A billion dollars in annualized revenue, reached in under 200 days from a standing start, tells you three things.
First, advertisers renew. A run rate built on one-time trial budgets does not compound like that. Second, OpenAI has an operational ad business now, with self-serve access through ads.openai.com for eligible businesses alongside the big agency partners like dentsu, WPP, and Publicis. Third, and this is the part I’d act on: when global agencies formalize their access, regional budget follows. The early auction in Southeast Asia is probably as cheap as it will ever be.
My honest read: this is a window, and it closes. The advertisers learning the surface this quarter are paying tuition at community-college rates. The ones starting next year pay full price, in auction terms, for the identical curriculum.
The second front: Grab, same week
The ads launch did not arrive alone. The same week, OpenAI and Grab announced “GO Forward with AI,” a two-year program training 30,000 driver, delivery, and merchant partners across six Southeast Asian countries, starting in Singapore and reaching Indonesia later this year.
The survey numbers inside that announcement are worth sitting with. Half of Grab’s surveyed Singapore driver and delivery partners already use AI tools, and 87 percent of those who don’t said they’re open to trying. That is mainstream adoption in the gig economy, not the tech-early-adopter demographic.
Put the two announcements next to each other and a strategy shows up: OpenAI is growing demand for ads (more usage, more markets) and supply of comfortable users (training, familiarity) in the same region at the same time. More daily ChatGPT habit in Indonesia and its neighbors means more inventory worth buying. Each launch makes the other one work harder.
What to do now
- Claim your Ads Manager access. Self-serve is open at ads.openai.com for eligible businesses. Even if you don’t spend yet, get the account live and learn the interface before your market gets crowded.
- Rebuild keyword research as question research. People talk to ChatGPT differently than they type into Google. Mine your customer service chats, WhatsApp threads, and sales calls for the literal questions people ask. Those questions, not your search terms, are your targeting vocabulary now.
- Run a small fixed-budget test for two to four weeks. The goal is learning, not performance. A budget you can lose without a meeting about it is the right size.
- Watch Shopee like a hawk. They ran the Brazil playbook first. Their category choices and creative formats in Indonesia are a free preview of what the surface rewards.
- Define success before launch, in writing. Click behavior from an answer surface will not match your Google Ads benchmarks. If you judge week-one numbers against search history, you will kill the channel before you understand it.
Three honest caveats
There are no public performance benchmarks for this region yet. Any agency selling you “proven ChatGPT Ads results in Indonesia” right now is selling you someone else’s guess. Measurement and attribution on the platform are young; treat every early number as directional rather than decisive. And if your buyers sit on Plus or Pro plans, they will never see your ads at all, which matters for pure B2B plays more than for consumer brands.
The bottom line
The mechanics here are the same ones we walked through when the US pilot went live earlier this month. What changed is the geography: the surface just arrived in a region where conversational commerce is already native behavior and mobile-first is the default, not the exception.
The cost of learning ChatGPT Ads in Southeast Asia is at its floor right now. That cost rises the moment agency budgets arrive at scale. A small test this quarter buys you something a large budget next year cannot: time on the surface before your competitors have any.
