Reality vs. Hype — No Technical Moat
@FT montage/Getty Images
Spending on infrastructure on the new Internet, back in the day, came in at $213B at its peak in 2000, right before the dotcom crash.
In the later 1990s, the telecom companies were saying that the internet demand was doubling every 100 days. OpenAI said in 2018 that demand for training is doubling every 3.5 months (about 100 days.)
Today, each new generation of Nvidia chips is roughly three times more powerful yet 50% more expensive. Accordingly, this means we will be able to 9X capacity by 2025 at 4X the cost — while demand is outpacing supply today, that won’t be the case for too long. So, if a path to profitability is not found by the time parity is reached, we will see GPU purchases reduced. Like in market times past, Nvidia stock will crash and take the market with it; GenAI will fade into reality.
It’s not only OpenAI creating enormous bogus valuations to fund themselves, it’s the space in general and the incumbent tech giants, too. All are heavily reliant on hype to play to their respective interests.
For example, if a company is generating $4 billion in revenue in 2024 but has $9 billion in cost, and with costs continuing to increase exponentially but with incremental revenue, with no sight of profitability on the horizon. What does logic and common sense tell you?
That company is OpenAI, of course, and it is also projecting $100 billion in revenue, without any substantiation or evidence, by 2029, just hype. It also threw out, for good measure, I guess, hitting profitability that same year. But Microsoft’s CFO, in their last quarterly earnings call, told us that they haven’t seen profitability from their OpenAI infrastructure investments and don’t expect any for at least 15 years. And even that came with disclaimers.
Open AI’s $157 billion valuation is based on an unproven, hubristic hypothesis, which fundamentally relies on its own AGI fantasy…future existence. Mainly to serve the priority of its own wealth objective.
OpenAI CEO Sam Altman: “Whether we burn $500 million a year or $5 billion — or $50 billion a year — I don’t care, I genuinely don’t. As long as we can figure out a way to pay the bills, we’re making AGI. It’s going to be expensive.” This sounds like burying your head in the sand to reality. It looks like Altman knows that his survival now rests on spending gobs of money, and maybe if he can make OpenAI too big to fail, it might save him.
OpenAI is just spewing bad economics and has created its own made-up rules outside of reality. Its made-up scaling laws, coupled with its own math, are only fooling the wilfully ignorant.
The economist and MIT professor Daron Acemoglu, and co-author of the book Power and Progress, Our 1000-Year Struggle Over Technology & Prosperity, recently wrote, “Given the focus and architecture of generative AI technology today… truly transformative changes won’t happen quickly, and few — if any — will likely occur within the next 10 years.”
OpenAI/GenAI is also enormously expensive to run, and the significant negative environmental impacts can’t be ignored. To keep these servers cool, it takes up ungodly amounts of energy. From best sources: we’ll likely need close to 190 GW of computing power (the capacity to train, fine-tune and execute algorithms) by 2030, with around 40% attributed to GenAI. Roughly, we are looking at 2 trillion dollars to be spent in the next 5 years. The climate will suffer, and something will have to give.
The big glaring problem for OpenAI is that it has no technical moat!
Eight of its eleven co-founders, including CTO Mira Murati and chief scientist Ilya Sutskever (who are launching competing ventures: Superintelligence), have left. When nearly all the talent that built your breakthrough technology goes elsewhere to build rivals, don’t you think it is worth asking why? As board chair Bret Taylor admits, we’re watching “the fastest technology commoditization cycle we’ve ever seen.”
GPT-4’s pricing per token has plummeted 98% since last year’s dev day. The gap between their SOTA models and open-source alternatives is narrowing with speed. Facebook is offering LLM/GenAI-based products/services, i.e., Llama — another unimpressive chatbot to crowd the market, for free, and Google has the overwhelming market share in Search — Gemini. Let’s face it: these products are all glorified search engines at the end of the day! So, from a long-term value and profitability perspective, for OpenAI, it’s in a commodity business and sector.
New technologies, no matter how amazing, don’t automatically translate into sustainable businesses. The reality is that the winners aren’t always those who built the most advanced and impressive technology or first-out-of-the-gate technologies — the winners are often those who can leverage a piece of technology and build a compelling ecosystem that is practical and useful to people. Real businesses.
It is also clear, from what we are hearing about the relationship with Microsoft, “What started as a lifeline — providing essential capital and computing resources — now risks becoming a constraint. Microsoft receives 75% of OpenAI’s profits until its $13B investment is recouped, followed by 49% until it hits $92B.” — Ashu Garg, Foundation Capital. That the honeymoon might be over.
Microsoft’s $650M acquisition of Inflection AI’s team recently looks very much like a hedge against overreliance on OpenAI. Microsoft didn’t invest billions in OpenAI because it loved the company; it did it because it didn’t want to miss out if OpenAI succeeded. Microsoft has so much money that it can do that, and that makes sense for a company like Microsoft to protect its market share and position itself to secure any future potential growth opportunities in that space.
OpenAI has created a phony ($157B) valuation, along with the fake math and hyperbolic pontification to ride the GenAI hype train. Eventually, real math and common sense will prevail.
Without a doubt, OpenAI can be credited for starting an “AI Revolution,” although I wouldn’t say that, let’s give it to Sam Altman anyway, his fragile ego needs it. ChatGPT is impressive. However, there is still no real, sustainable, profitable business story that they can credibly articulate against the prevailing backdrop of business realities. Starting a revolution is a far cry from profiting from it.
With the ever-increasing commoditization of the sector, there is no reason to bet on “OpenAI” even being around similar to present form over the long run.
The most useful (and valuable) companies to humanity in the AI space do not authentically exist yet, but they’ll be the ones not driven by technology hype, harnessing AI’s potential instead, purposefully, to solve very specific problems. Not engaged in exaggerations about its made-up scaling laws about the capability of LLMs and one-day achieving fantasy AGI. Which only serves to drive abstraction and hallucination, holding back real progress and humanity.
Comments