The World is Ads
Here we go again, the tech press is having another AI doom cycle.
I've primarily written this as a response to an NYT analyst painting a completely unsubstantiated, baseless, speculative, outrageous, EGREGIOUS, preposterous "grim picture" on OpenAI going bust.
Mate come on. OpenAI is not dying, they're not running out of money. Yes, they're creating possibly the craziest circular economy and defying every economics law since Adam Smith published 'The Wealth of Nations'. $1T in commitments is genuinely insane. But I doubt they're looking to be acquired; honestly by who? you don't raise $40 BILLION at $260 BILLION VALUATION to get acquired. It's all for the $1T IPO.
But it seems that the pinnacle of human intelligence: the greatest, smartest, brightest minds have all come together to... build us another ad engine. What happened to superintelligence and AGI?
See if OpenAI was not a direct threat to the current ad giants would Google be advertising Gemini every chance they get? Don't forget they're also capitalising on their brand new high-intent ad funnel by launching ads on Gemini and AI overview.
Let's crunch the numbers.
Quick Recap of OpenAI's 2025
March: Closed $40B funding round at $260B valuation, the largest raise by a private tech company on record.
June: Hit $10B ARR.
July: First $1B revenue month, doubled from $500M monthly in January.
November: Sam Altman says OpenAI expects $20B ARR for 2025.
Reached 800M WAU, ~190M DAU, 35M paying subscribers, 1M business customers.
January 2026: "Both our Weekly Active User (WAU) and Daily Active User (DAU) figures continue to produce all-time-highs (Jan 14 was the highest, Jan 13 was the second highest, etc.)"
January 16, 2026: Announced ads in ChatGPT free and Go tiers.
Yes, OpenAI is burning $8-12B in 2025. Compute infrastructure is obviously not cheap when serving 190M people daily.
Predicting OpenAI's Ad Strategy
So let's try to model their expected ARPU (annual revenue per user) by understanding what OpenAI is actually building and how it compares to existing ad platforms.
The ad products they've confirmed thus far:
- Ads at bottom of answers when there's a relevant sponsored product or service based on your current conversation
Rollout:
- Q1 2026: Limited beta with select advertisers
- Q2-Q3 2026: Expanded to ChatGPT Search for free-tier users
- Q4 2026: Sidebar sponsored content + affiliate features
- 2027: Full international expansion, self-serve platform
Testing starts "in the coming weeks" for logged-in adults in the U.S. on free and Go tiers. Ads will be "clearly labeled and separated from the organic answer." Users can learn why they're seeing an ad or dismiss it.
Their principles:
- Answer independence: Ads don't influence ChatGPT's answers
- Conversation privacy: Conversations stay private from advertisers, data never sold
- Choice and control: Users can turn off personalization and clear ad data
- Plus, Pro, Business, and Enterprise tiers won't have ads
They also mentioned a possibility of conversational ads where you can ask follow-up questions about products directly.
Revenue targets: Reports suggest OpenAI is targeting $1B in ad revenue for 2026, scaling to $25B by 2029, though OpenAI hasn't confirmed these numbers publicly. We can use these as the conservative benchmark, but knowing the sheer product talent at OpenAI, the funding and hunger. I think they're blow past this.
Personal speculations on integration:
- Self-serve platform: Advertisers bid for placements, super super super likely, exactly what Google does, probably their biggest revenue stream.
- Affiliate commissions: Built-in checkouts so users can buy products inside ChatGPT, OpenAI takes commission, similar to their Shopify collab.
- Sidebar sponsored content: When users ask about topics with market potential, sponsored info appears in a sidebar marked "Sponsored"
Now let's compare this to existing ad platforms:
Google: Intent + Vertical Integration = Highest Revenue
- How it works: Auction-based system where advertisers bid on keywords. Ads appear in search results based on bid + quality score.
- Why it works: High intent (search queries) + owns the entire vertical stack (ad tech, auction system, targeting, decades of optimization)
- Ad revenue: [$212.4B in ad revenue in the first 3 quarters of 2025]https://www.demandsage.com/google-ads-statistics/ (8.4% growth from 2024's $273.4B)
- Google doesn't report ARPU so we need to calculate it: ARPU = $296.2B (projected) ÷ 5.01B = $59.12 per user annually.
Meta: No Intent + Vertical Integration = High ARPU
- How it works: Newsfeed ads delivered via auction. Meta's Andromeda AI evaluates bid + predicted action rate + ad quality to determine placement.
- Why it works: Passive scrolling = low purchase intent, but on a massive scale + owns targeting infrastructure + Andromeda AI
- ARPU: $68.44 in North America, $49.63 globally (Q1 2025)
- Revenue: $160B in 2024 (97.3% of total revenue)
Twitter/X: Engagement + No Vertical Stack = Low ARPU
- How it works: Auction-based promoted tweets in timeline. Advertisers only pay when users complete actions (click, follow, engage).
- Why it works: Timeline engagement, CPC ~$0.18, but doesn't own vertical stack and does it on a smaller scale
- ARPU: ~$5.54 ($2.3B revenue ÷ 415M MAU)
- Revenue: ~$2.3B in 2025
ChatGPT: High Intent + No Vertical Stack = Where Does It Sit?
- Intent level: High. 2.5B prompts daily includes product research, recommendations, comparisons. More intent than Meta's passive scrolling, comparable to Google search.
- Vertical integration: None. Yet.
- Scale: 1B WAU by Feb 2026, but free users only (~950M at 95% free tier).
So where should ChatGPT's ARPU sit?
It sits with Search, not Social.
Which puts it between X ($5.54) and Meta ($49.63). OpenAI has better intent than Meta but worse infrastructure. They have more scale than X but no vertical integration. When a user asks ChatGPT "Help me plan a 5-day trip to Kyoto" or "Best CRM for small business," that is High Intent. That is a Google-level query, not a Facebook-level scroll.
We already have a benchmark for this: Perplexity.
In late 2024/2025, reports confirmed Perplexity was charging CPMs exceeding $50. This is comparable to premium video or high-end search, and miles above the ~$2-6 CPMs seen on social feeds.
If Perplexity can command $50+ CPMs with a smaller user base, OpenAI’s "High Agency" product team will likely floor their pricing there.
Super Bullish Target ARPU Trajectory:
- 2026: $5.50 (The "Perplexity Floor") - Even with a clumsy beta and low fill rate, high-intent queries command premium pricing. If they serve just one ad every 20 queries at a Perplexity-level CPM, they hit this number effortlessly.
- 2027: $18.00 - The launch of a self-serve ad manager (like Meta/Google) allows millions of SMBs to bid. Competition drives price.
- 2028: $30.00 - This is where "Ads" become "Actions." OpenAI won't just show an ad for a flight; they will book it. Taking a cut of the transaction (CPA model) yields 10x the revenue of showing a banner.
- 2029: $50.00 (Suuuuuuuper bullish case) - Approaching Google’s ~$60 ARPU. By now, the infrastructure is mature, and "Conversational Commerce" is the standard. This is what Softbank is praying will happen.
And we're forgetting that OpenAI have a serious serious product team, I don't doubt for once they'll be fully capable of building out the stack and integrating ads til they occupy your entire subconscious.
In fact they hired Fidji Simo as their "CEO of Applications", a newly created role that puts her in charge of their entire revenue engine. Fidji is a Meta powerhouse who spent a decade at Facebook working on the Facebook App and... ads:
Leading Monetization of the Facebook App, with a focus on mobile advertising that represents the vast majority of Facebook's revenue. Launched new ad products such as Video Ads, Lead Ads, Instant Experiences, Carousel ads, etc.
Launched and grew video advertising to be a large portion of Facebook's revenue.
Being Realistic About Competition
ChatGPT will hit 1B WAU by February 2026.
But 1.5-1.8B free users by 2028? That assumes zero competition impact from anyone, certainly not the looming giant Gemini. Unrealistic.
Let's estimate growth super conservatively accounting for competition:
- 2026: 950M free users (1B WAU × 95% free tier)
- 2027: 1.1B free users (slower growth as market saturates)
- 2028: 1.2-1.3B free users (competition from Google, Claude)
- 2029: 1.4B free users (mature market, multi-player landscape)
The main revenue growth comes from ARPU scaling not just user growth.
Predicting 2026
Crunching all the numbers from "High Intent" model, 2026 looks different.
Base revenue from subscriptions + enterprise + API: $25-30B
- 35M paying subscribers: $8.4B minimum (conservatively assuming all at $20/mo Plus tier)
- Definitely higher with Pro ($200/mo) and Enterprise (custom pricing)
- Enterprise/API: $2.3B in 2025 → $17.4B by mid-2027
Ad revenue (year 1): ~$5.2B
- 950M free users x $5.50 ARPU
- ChatGPT does 2.5B prompts daily this is what advertisers would class as both higher engagement and higher intent than passive scrolling (although you can fit more ads in a scroll than a chat)
- Reality Check: This assumes they monetise typical search queries at rates Perplexity has already proven possible.
Total 2026 Revenue: ~$30-35B.
Projecting 2027-2029
These projections use futuresearch.ai's base forecast ($39B median for mid-2027, no ads) + advertising overlay from internal OpenAI docs + conservative user growth.
2027:
- Base revenue (no ads): $39B
- Ad revenue: $19.8B (1.1B free users × $18 ARPU)
- Total: $58.8B
2028:
- Base revenue (no ads): $55-60B
- Ad revenue: $36-39B (1.2-1.3B free users × $30 ARPU)
- Total: $91-99B
2029:
- Base revenue (no ads): $70-80B
- Ad revenue: $70B (1.4B free users × $50 ARPU)
- Total: $140-150B
The World is Ads
Ads were the key to unlocking profitability, you must've seen it coming, thanks to you not skipping that 3 minute health insurance ad - you, yes you helped us achieve AGI!
Mission alignment: Our mission is to ensure AGI benefits all of humanity; our pursuit of advertising is always in support of that mission and making AI more accessible.
The A in AGI stands for Ads! It's all ads!! Ads that you can't even block because they are BAKED into the streamed probabilistic word selector purposefully skewed to output the highest bidder's marketing copy.
Look on the bright side, if they're turning to ads it likely means AGI is not on the horizon. Your job is safe!
It's 4:41AM in London, I'm knackered. Idek if I'm gonna post this because I love AI and do agree that some things are a necessary evil to achieve a greater goal (AGI). Nevertheless, if you have any questions or comments, shout me -> ossamachaib.cs@gmail.com.