Recently, it was revealed that OpenAI is expected to record $1.6 billion in annualised revenue in 2024. This is a 300 million bump up from the predicted $1.3 billion in October last year, coming from ChatGPT subscriptions, API access and others.
Meanwhile, competitor Anthropic expects at least $850 million in annualised revenue in 2024 up from $100 million three months ago, which they had informed the investors about. Anthropic’s projected revenue indicates a potential surge to $70 million per month by the following year, driven by Claude, its text-generating AI, serving customers like Notion, Quora, and DuckDuckGo.
On the other hand, bootstrapped AI image-generator company Midjourney generated a revenue of $200 million in 2023 from its subscription model that charges users between $10 and $120 monthly.
Is it Sustainable?
As these companies compete with each other in the space, the potential growth and evolution of their businesses, relative to their current capabilities, remains uncertain for various reasons. For instance, the never-ending need to raise substantial capital, with CEO Sam Altman indicating that OpenAI potentially needs $100 billion to run the business, while the financial losses mount upto $540 million.
Plus, big tech companies like Microsoft, Google, and AWS which are injecting capital also get a share of the generated revenue alongside computational cost and accelerate research and development to keep up with the competition.
Additionally, numerous prominent figures in the industry are now expressing doubts over whether current models will sustain the pace of advancement or have they already hit a wall.
Echoing a similar sentiment, Bill Gates expressed doubts about the progress of generative AI saying that it has plateaued. And despite the optimistic view from some at OpenAI about GPT-5’s potential, Gates believes current generative AI has hit a ceiling. While acknowledging the impressive leap from GPT-2 to GPT-4, he admitted the possibility of being wrong about its limitations.
Ashray Malhotra, co-founder and CEO of Rephrase.ai, also shared his thoughts in a recent podcast on the cost of further innovation and the scale of it.
“GPT 5, whenever they train it, [it turns out] like 10 times more expensive and only 10% better than GPT 4 and you realise that for the next big thing, you need a fundamentally different architecture than a diffusion or transformers. Then you’re back in the wilderness, like, what is that next big thing?”
For How long?
Insiders suggest that Anthropic is anticipated to potentially hit $1 billion in annualised revenue next year, around $83 million monthly. Contrarily, projections expect OpenAI to generate $5 billion in annualised revenue by 2024.
However, while the revenue grows and the company continues to raise endlessly from investors and spend money on research and development, it would still keep them a long way from profitability.
How are They Doing it?
To deal with revenue issues, companies will need to look at diversifying their businesses and building in-house capabilities. This would reduce reliance on other companies, cut costs and bring in a newer stream of revenues.
OpenAI, for instance, is moving ahead of just a subscription model and catering to enterprises. It is also said to be venturing into the AI server chip business for which Altman is raising money from Abu Dhabi firm, G42.
Others like Stability AI, despite earlier revenue estimates exceeding $10 million annually, are changing gears from open-source by introducing a subscription fee for enterprise customers. The move towards subscriptions comes as Stability’s open-source strategy failed to yield significant income.
A focus on increasing and projecting safety and reliability in the technology could also lead to applications across industries like healthcare, marketing, banking, telecom, defence and others. This would mean a strong, steady and bigger stream of revenue.
Alternatively, one could stick to Midjourney co-founder David Holz’s philosophy, which stands as a testament to running a profitable bootstrapped company with sustainable growth. He owes it to an excellent product focussed on catering to a specific customer base without raising any capital from VCs.
Perplexity CEO, Aravind Srinivas also recently justified their lack of a focussed strategy as it tries to enter the search business and others. He posted on X saying that “Jeff Bezos invested in Google in 1998, before it had any business model figured out or was sky-rocketing in search traffic. His investment in Perplexity is symbolically significant.”
