Ed Zitron on his Where's Your Ed At blog: There Is No AI Revolution
I kept coming back to one thought: where's the money?
No, really, where is it? Where is the money that this supposedly revolutionary, world-changing industry is making, and will make?
The answer is simple: I do not believe it exists.
I was recommended this article recently after posting about how I used an AI tool for something, and it’s lengthy but I gave it a read. I found the piece kind of compelling, and it’s 9,000 word count can be summed up as basically: OpenAI isn’t profitable and their market share is so far ahead of everyone else right now, both of which prove that AI is not a real industry. I guess I’d classify this article as one that I agree with the general argument, but the author loses me on many of the details.
Below is a lengthy quote, but I think it hits at the core argument of the author:
The only product that OpenAI has succeeded in scaling to the mass market is the free version of ChatGPT, which loses the company money with every prompt. This scale isn't a result of any kind of product-market fit. It's entirely media-driven, with reporters making "ChatGPT" synonymous with "artificial intelligence."
• As a result, I do not believe that generative AI is a "real" industry — which I define as one with multiple competitive companies with sustainable revenue streams and meaningful products with actual market penetration — because it is entirely subsidized by a combination of venture capital and hyperscaler cloud credits.
• ChatGPT is popular because it is the only well-known product, one that's mentioned in basically every article on artificial intelligence. If this were a "real" industry, other competitors would have similar scale — especially those run by hyperscalers — but as I'll get to later, data suggests that OpenAI is the only company with any significant user base in the entire generative AI industry, and it is still wildly unprofitable and unsustainable.
On the one hand, I respect the audacity to reference the quickest growing new product in human history which has only grown is usage since that initial spike in 2022 as “not mass market”. I also find it remarkable to say ChatGPT, a name literally no one knew 3 years ago, is only popular because people have heard of it and that only “real” industries have numerous competitors of similar scale. I guess search is not real since Google has a much bigger market share than OpenAI, computer operating systems aren’t a real industry since Microsoft has higher market share than OpenAI, and I guess online shopping isn’t real since Amazon dominates as well. Amazon is a particularly interesting one since they were famously unprofitable (albeit to a lesser extent than OpenAI) for many years before they finally started turning a profit.
I personally think that AI tools are practical for numerous use cases and they are valuable additions to what we can do with computers (I literally wrote about this earlier today). I do agree that the spending that companies like OpenAI are doing today are unsustainable, and they are a result of a new market racing to get ahead now so that they can be the established option in the market long term. Like the author, I agree that the moat for these companies is not very strong, as every innovation we see in the world of LLMs is almost immediately adopted by every other major competitor out there. I also agree that the grandiose statements from AI executives is eye-rolling.
Where I disagree with the author is the assumption that these costs will continue forever and only increase. I think we’re seeing that LLMs are reaching the point of diminishing returns, as we’re seen from recent updates like Claude 3.7 and GPT 4.5. OpenAI spent more time and money on building GPT 4.5 than any previous model, and yet even AI enthusiasts are struggling to see how it’s meaningfully better than 4o that it’s replacing. To me, this indicates that models of today’s quality will become cheaper, faster, and less energy intensive to run in time, bringing all of these costs down. I’m sure there will be groups continuing to try to push the bleeding edge forward, but I expect these efforts to become more niche as what we have is good enough for most things most people need them to do. Soon enough we’ll have models of the quality of GPT 4.5 and Claude 3.7 running locally on our smartphones – there may be technically better models out there on high-intensity servers, but unless something radically changes, I expect those to become the exception, not the rule. Let’s not forget that CG graphics for movies like Toy Story required server farms to run at full tilt for days to render individual shots. Today, a consumer laptop can render massively more complex scenes in far less time.
Also, as someone who has written about the intentionally misleading reporting on LLM electricity usage several times in the past year, this line from early on caught my eye:
OpenAI's Sam Altman and Anthropic's Dario Amodei have used a compliant media and braindead investors to frame unprofitable, unsustainable, environmentally-damaging and mediocre cloud software as some sort of powerful, futuristic automation.
You may not be surprised to know that neither in this article or the linked article in the above quote does the author reference any numbers in regards to energy consumption, it’s just assumed to be horrendous. I’ll say it again, I do think that LLM usage should be more efficient, I do think energy use will decline as “good enough” models become easier to run at low energy costs, but I also think that these damages are overstated by many.
So is AI an “industry”? I think there is a compelling case to be made that OpenAI, Anthropic, and the like are spending tons of money to bring a new technology to market where platform holders will copy their work for a fraction of the cost and embed tools in their software. Maybe as Steve Jobs once said about Dropbox, LLMs are features, not products. I do think it’s important to add that Dropbox and numerous file sharing competitors are still around and are profitable businesses that have found a niche in the market they can serve.