The AI re-alignment is here
These latest news articles reminded me why this newsletter is called ‘Recurrent Patterns’, which on occasion can be called ‘Here we go again’.
The AI world is getting ‘loopy’ (TechCrunch)
Companies are scrambling to stop employees from maxing out AI budgets with small tasks (TechCrunch)
OpenAI’s financials have leaked, showing $21 billion in losses against $13 billion in revenue (Fortune)
OpenAI Considers Drastic Price Cuts, Anticipating War for Users With Anthropic (The Wall Street Journal)
SpaceX stole the IPO thunder from both OpenAI and Anthropic. SpaceX has at least one division or line of business (Starlink) which is profitable, unlike OpenAI or Anthropic. Based on the article from Fortune, OpenAI is losing money at ludicrous speed and one can extrapolate that Anthropic is not far behind.
As both companies are preparing for their respective IPOs they want to make their books look as good as possible.
First, let’s get the revenue up. The price increase went unnoticed until a few months later when the bills started arriving at the CFO offices or the CTO was called into the CFO office to answer uncomfortable questions about the year’s budget being spent within Q1. The price hike got lots of pushback from the customers where the business started asking the question: What are we getting in return? Uber is one of these companies. Now OpenAI and Anthropic have to decide which thing they are willing to sacrifice — revenue or market share. Neither looks good when you are preparing for the IPO.
Side note: Despite acquiring more customers, OpenAI is losing market share to its rivals. Not a good story for the IPO.
That brings us to another interesting conversation. Neither company is remotely profitable. Based on the leaked document, OpenAI is losing money with every additional customer. I don’t think they can make up the difference in volume?
The one part which the article from Fortune didn’t address is the marketing spend by OpenAI. The line reads — “Sales and Marketing: $5.73 billion (2025) up from $1.11 billion (2024).” Is $6 billion too much, too little, does it make sense? For comparison, here are the top 25 advertisers in the world. Based on these numbers, OpenAI would be number 14. Perhaps you can argue that the $5.73 billion was not only marketing but sales as well. Let’s assume that the sales team was $1 billion … In that case OpenAI would drop to number 18 but still would be ahead of McDonald’s, Toyota or Coca-Cola.
In marketing, to assess the success of campaign reach is to ask people to recall an ad by a brand. Close your eyes and tell me which ads you can recall from the last 24 hours. I don’t think it is OpenAI. Another data point for you to consider is the ROI on marketing spend. The worst performer from the top 25 is L’Oreal with 2.9x. That company spent $13.1 billion and was able to attribute almost 3x revenue in return. OpenAI reported $13 billion in revenue. Either the marketing spend hides other expenses to make the numbers look better or they should fire their CMO.
The second thing which both companies have to do is to keep the hype up. What else can you say about your technology to make the market believe that it is a) awesome and b) there is more you can do with it.
Well, we witnessed over-selling the threat of Claude Fable 5 (and Mythos by Anthropic which resulted in a ban by the US government aka free publicity). And the introduction of ‘loops’ where “agents are prompting agents that then write the code,” and “The loop takes it a step further by authorizing a swarm of agents to work continuously in the background, endlessly.”
What would be the advantage of an ever-changing code base that changes in totally unpredictable ways? Using as many tokens which will result in infinite token charges.
Except, who has an infinite budget? Before any budget considerations, companies would punish people for not using enough tokens. But after price increase, the same companies are instituting tokens rationing.
So here you have it. Unprofitable companies with ‘cool’ and immature technology, weak business models, no pricing power, vaguely defined products. Companies which are trying to cash out through public markets. Their commitments to spend even more billions on new data centers is a dream.
The recurrent pattern? The disconnect between the tech and business people. With every new technology, the geeks come out and tell the business that this technology for sure will deliver riches beyond any imagination. The business is always skeptical but eventually gives in and starts spending money to reach the magical land of Neverland. Now, the business is waking up to the unrealistic promises and the geeks are running out of ideas and money.