Can AI Become Profitable?

The Matrix, Skynet, Master Control Program. These are the nefarious villains humanity's most pessimistic views of AI have created.

Following social media chatter makes it hard to fathom a world without omnipotent and omniscient AIs dominating humanity, if the alarmists are to be believed. Already, it would seem, AI is making humanity redundant by performing skilled and unskilled labor humans have been needed for.

Baked into the belief AI will profoundly change the world is a presumption: the companies providing the AI will become massively profitable. The billions of dollars continuing to be invested in those companies demonstrate this anticipation.

But is this presumption grounded in reality?

For all of AI's promises, there are many headwinds facing the industry and technology itself.

Apple's AI development has underwhelmed. A sentiment gaining traction is that they have lost the AI arms race, as evidenced by their reveal of Liquid Glass, a new, immersive mobile experience.

Meta has just invested a tremendous amount of money into creating an AI brain trust and acquiring Scale AI, a data labeling and model evaluation company. But rumor has it Zuck is concerned Meta isn't making the kind of progress he believes it needs to become the top AI company, making this behavior seem like he's chasing a sunk cost.

The latest update in AI melodrama is the decaying relationship between Microsoft and OpenAI, the flagship AI company. An acrimonious divorce is unfolding in real time.

The two companies have been in talks to transform OpenAI from a nonprofit to a for-profit entity, which risks diluting Microsoft's ownership stake. Apparently, Microsoft does not like where the discussions have been heading and is strongly considering terminating them and continuing the current commercial contract to maintain access to OpenAI's technology until 2030.

OpenAI is accusing Microsoft of anticompetitive behavior in what looks to be a gambit to free itself from dependency on Microsoft.

So, all is not harmonious in the AI world. Far from it. An important question is: why does OpenAI want to change its legal structure? The answer is an obvious and timeless one: it wants more money. Specifically, it needs to raise more investment capital. Additionally, it does want to expand to other partners, for instance, hosting their software on AWS in addition to Azure.

That first answer underscores a profound challenge OpenAI and other leading AI companies face with the technology. The investments are incredibly expensive, and the path toward profitability is the actual point of these endeavors. No one's buying the insipid declarations of noble intentions to benefit humanity tech CEOs make.

Though OpenAI showed impressive revenue growth last year, their 2024 revenue was ~$10B; the company lost an estimated $5B in the same period because of the exorbitant costs it incurs.

Turns out making and providing advanced AI is expensive. The operating costs are staggering thanks to requiring the best and brightest tech talent, unprecedented amounts of compute power (keep thanking ChatGPT for its outputs everyone!), data acquisition (and this is without paying what those data creators consider to be fair), and of course, research and development.

That last part is particularly noteworthy. The atom bomb of the AI arms race is Artificial General Intelligence (AGI) , that is, machines that can think like, and possess the awareness of, human beings. Every AI company understands that whoever makes this breakthrough wins, making the competition irrelevant.

This creates a potentially catastrophic problem. What if AGI can't be achieved? We hardly understand human consciousness as it is. How consciousness arose is one of the most hotly debated topics among a plethora of scientific and philosophical disciplines. In other words, we don't have an answer.

The "how" of creating AGI is thus a mystery the most sanguine takes on the development of AI haven't proven are any closer to being answered, despite the bloviating techno-babble they use to obscure this reality.

I don't know how long AI pioneers, and their investors, can continue to subsidize the development of AGI, but it's certainly not indefinitely. If AGI can't be achieved, when do tech giants throw in the eventual towel on it? More importantly, what will they do if and when that reality sets in? Will the business model adapt to become profitable by providing the non-AGI uses it's capable of?

I remain pessimistic. The unceasing drive towards AGI indicates tech giants are betting their futures on it, and when it becomes clear it won't happen, there will be a reckoning.

One thing is for certain if that ends up being the case, the doomsday predictions of AI obsoleting humanity will not come to pass, so there's at least one reason to remain optimistic. The future will likely lack machine overlords, replete instead with em dashes.