The AI reality check

As the proverb goes — the writing is on the wall for OpenAI.

Let’s start with the (investor) positive news:

• Alphabet revenue tops expectations on record quarter for cloud unit

• Amazon tops cloud expectations on strong AI demand, shares rise

• Microsoft expects strong cloud business growth, plans record capital spending

Google, Amazon and Microsoft are doing just great. Who is not doing great?

• OpenAI’s revenue, growth estimates fall short as company races toward IPO

• Oracle, CoreWeave lead AI selloff on OpenAI growth concerns

And then you see another headlines:

• OpenAI brings its models to Amazon’s cloud after ending exclusivity with Microsoft

• OpenAI and Microsoft loosen ties in revised AI deal

True, OpenAI closed a $122 billion funding round where Amazon and Microsoft participated, but that money was swiftly allocated to the compute resources with these respective funders.

OpenAI is losing steam. As per the above report, not only is the revenue not climbing fast enough, more importantly, the user growth is stagnant. And without these two things it is difficult to raise more money and especially to prepare for the long expected IPO. Not growing your customer base, while your competition is adding them, also means that OpenAI is losing market share.

But just maybe, OpenAI hit the technology adoption curve where the early enthusiasts were willing to experiment, try, get excited — why not, it was free.

I know, having 900 million users on your platform is impressive, except the vast majority are free loaders and despite the various attempts, very few are converting to paid customers. Despite the recent introduction of advertising on the free accounts, that revenue won’t be able to deliver the stellar financial windfall OpenAI hoped for.

The quality of these ads — despite ChatGPT being part of the conversation — is somehow questionable. When a user was planning a trip to Beijing, the ad presented was for Chinese food delivery.

However, the use and adoption of AI is going up. It must be when we are experiencing this crazy trend — tokenmaxxing, where developers are trying to outcompete each other for who uses AI more for code development. The anecdotes suggest that part of the compensation package for new hires is how many tokens (a consumption unit of AI) the developer will be able to use. Or we hear that developers are let go and the remaining have to use AI to keep things going.

One could easily pick the other stories questioning the use and maturity of the technology, like “A founder says Cursor’s AI agent deleted his startup’s database, causing chaos for customers” which could be described and dismissed as unlucky.

Who cares about a tiny company? But when you see this headline — Nvidia VP Says AI Costs ‘Far’ More Than Human Employees — you pause and wonder when you read “AI compute now costs more than the employees using it, making AI more expensive than human labor.” It is coming from a $5 trillion market cap company. A company which is fueling the AI market with both chips and money.

Bryan Catanzaro from Nvidia is not alone. You also get Uber’s chief technology officer, Praveen Neppalli Naga, sharing that the company’s shift toward AI coding tools is driving up costs. “I’m back to the drawing board because the budget I thought I would need is blown away already,” he said.

This is the recurrent pattern. After the initial awe and experimentation with the new technology, comes a moment when the technology has to mature and deliver real value, not the marketing hype. The mature vendors get it. The re-alignment is coming.