TL;DR?
As I've described in earlier articles, today's “AI world” rests primarily on four core premises:
- LLMs are a useful but immature technical innovation. They are not “intelligent” in any strict sense, but they offer two decisive advantages: first, the ability to extract information from large volumes of unstructured data, and second, a natural language interface that simulates human interaction and makes complex systems far easier to use.
- Their roots lie in social IT. The primary business model there is monetising user data and attention. In business IT, that has traditionally been taboo, which is why business IT pays its way through classic sales and licensing models.
- AI is a feature, not a product. Artificial intelligence makes less sense as a standalone product and far more sense as a feature built into existing systems and processes.
- LLMs are extremely resource-hungry. Both development and operation demand enormous computing capacity and energy. Any business model has to account for that.
These premises explain why the AI industry is still standing on shaky legs. The ongoing hype has triggered massive inflows of capital, yet the return on investment (ROI) of many models remains unclear. At the same time, this very fresh capital could be an opportunity: the chance to transition to more sustainable business models.
OpenAI sets the pace
The current top dog, OpenAI, largely sets the direction for the industry. The company is trying to gain a foothold in as many segments as possible:
- Social IT: With the standalone app Sora-2, OpenAI wants to establish a classic social feed model, much like what we know from Instagram, TikTok and the like. It will be interesting to see how many monthly active users OpenAI can pull away from those platforms and how the business model, presumably ad-based, develops. It's also worth keeping an eye on what OpenAI has in mind for the browser market.
- Business IT: Here OpenAI is pursuing several strategies:
- Its own workplace software suite is meant to let it sell its LLMs inside business software with a traditional look and feel.
- A recruiting platform is meant to shake up an HR market that AI has already turned upside down.
- OpenAI is conspicuously quiet in the “generative software” market right now, one of the most dynamic and profitable AI segments. Anthropic, with its latest Claude versions, still holds the lead here and is the favourite of many developers. But OpenAI has already tried to buy the AI coding tool Windsurf and has its own iron in the fire with Codex. Competition is especially fierce here, because the high API costs of Anthropic and OpenAI can pay for themselves quickly once you set them against the person-days saved on consultants and developers (an article on that is coming soon, stay tuned).
- Hardware: OpenAI is opening new fronts here too, both in processors with its own chips and in the possible development of consumer hardware, for which the company is said to have poached numerous former Apple employees.
From a business perspective, these strategies are genuinely interesting. OpenAI is trying to mutate from a start-up into a “conglomerate of start-ups” (understandable, if you know Sam Altman's track record). That could prove to be a smart move. It can also look rather reckless to take on established giants like Meta, Alphabet or Nvidia.
Fact is: simply because AI is a feature and not a product, this market will split in two:
- Pure AI vendors who build products around their AI
- Existing product vendors who build AI into their products
Both will compete hard for the same market share, so what decides it is what has always decided it: whoever does better product marketing, sells better and offers better customer service will win. Seen that way: nothing really new under the sun, just old wine in new bottles 😉
Between hype and reality
What keeps blocking the view: US financial markets are increasingly driven by hype, and loudmouths like Sam Altman, Elon Musk et al. keep bluffing with grand announcements to steer capital into their coffers. This stopped having anything to do with sustainable investing a long time ago, and the enthusiasm they stoke can very quickly turn out to be a flash in the pan. Tesla is a good example: a skyscraper thrown up in a hurry that is now being attacked by two bulldozers, the established carmakers on one side and the new, rising Chinese all-rounders on the other.
The same could happen to OpenAI.
