After the Democrats won the mid-terms the anti-AI rhetoric has strengthened and has gone beyond reason. The single biggest slogan is - ‘AI is a bubble’. Here are my reasons why it is very likely not
Claims:
- AI is Fake - No revenue now. All $s in future promises and revenues are speculative
- AI is a $ Fraud - Like Enron, i.e. mis-representation and books are cooked etc.
- AI is a tech Fraud - like Theranos. The technology doesn’t work or its effectiveness is vastly overstated
Rebuttal:
AI is Fake
The spend is real money. It is funded mostly by the cashflow of Big tech and hyperscalers. The chipmakers are actually showing real revenues and net income. Some of them are even buying back their stock in a way to return $s to share-holders The spenders like Frontier labs like Anthropic and OpenAI have at least 200B$s in ARR between them and its growing.
Peak spend argument - This is true because it is not mathematically possible to grow in the second derivative forever. i.e. the $d^2(spend)/dt^2$ might become 0 or even turn negative. But the spending rate $d(spend)/dt$ will be at this new elevated level given very long-term contracts for pretty much until the end of the decade. Every time there a new usecase, a new capability, a break-through, an incremental progress is announced, people will realize how much under-allocated they have been all along.
The only remaining open questions (non-trivial) are:
- For Chip, interconnects, neo-clouds, hyperscalers - How much longer will the spending last? Is it cyclical like all other cycles previously?
- For everyone in the AI business including frontier labs - What is the TAM for AI models across all possible usecases?
AI is a $ Fraud
While there are companies that are trying to grift off of AI hype like Oracle and Dell. Essentially dinos trying to do a hand-stand. The large and consequential companies have very visible revenues and most of them are public companies with several alternate streams of revenues.
The only valid argument on the financial front is the matching of the spend rate with the revenue rate. This is an operational detail but not trivial. Several firms in the past have gone bakrupt just for this seemingly ’trivial’ reason. This risk is real and it remains. Especially true for under capitalized, non-diversified spenders.
AI is a tech Fraud
This is where I would focus on. I started my career as an engineer and have known coding from the micro-processor level, writing hex-code by hand all the way to Object oriented programming and system architecture. I have been using AI coding assistants for a while now and I can attest that the improvements are very real.
At least for Software Engineering as a profession the productivity gains are indisputably tangible. The rate of adoption within the Software Engineering community is nearly 100%. This is not anencdotal.Forums like Stack-overflow, the previously go-to site for all Software related questions, are experiencing extinction level drops in user engagement. Not using AI for coding now is as odd as being an Amish zealot.
Finally, chanting the anti-AI slogan is a disservice to humanity as it can be immensely helpful in areas like healthcare and development. Especially in under-developed and developing countries. The cures to carious diseases, better governance, more empowerement are all hidden in the weights of these models.
AI is not a bubble.