Lowering the Boom on the New Boom Times
Good timing on my column for members of The Inner Ring yesterday entitled: Almost All A.I. Investments are Going to Zero. Just one day later we get some numbers to quantify just how crazy it is out there right now:
Investors poured $27.1 billion into A.I. start-ups in the United States from April to June, accounting for nearly half of all U.S. start-up funding in that period, according to PitchBook, which tracks start-ups. In total, U.S. start-ups raised $56 billion, up 57 percent from a year earlier and the highest three-month haul in two years.
A.I. companies are attracting huge rounds of funding reminiscent of 2021, when low interest rates and pandemic growth pushed investors to take risks on tech investments.
Boom times are back! But per the numbers, this one is almost entirely driven by one sector. And startups not in that sector are still having a hard time raising. Which is a problem because it's pushing more companies unnaturally into that sector. This is, of course, how bubbles form. That doesn't have to be a bad thing necessarily, but per my post, the unique market dynamics at the moment will unfortunately make it a bad thing for almost all of these companies.
And actually, I believe we're already seeing the signs that we need more companies built to solve problems where AI is a tool they use, versus companies being built with AI as the starting point and trying to figure out the pain points they wish to solve with it.
An analysis of 125 A.I. start-ups by Kruze Consulting, an accounting and tax advisory firm, showed that the companies spent an average of 22 percent of their expenses on computing costs in the first three months of the year — more than double the 10 percent spent by non-A.I. software companies in the same period.
“No wonder V.C.s are throwing money into these companies,” said Healy Jones, Kruze’s vice president of financial strategy. While A.I. start-ups are growing faster than other start-ups, he said, “they clearly need the money.”
This is an important point. Unlike the last investing frenzy – just a few short years ago! – this isn't money being thrown at companies for nothing. I mean, sure, some of it is. But much of it is because the cost of compute with A.I. is so expensive that these companies are burning through capital at high rates. And, oddly, the biggest beneficiaries of that spend are either the tech giants with cloud A.I. solutions (Microsoft, Google, and Amazon), the LLM players (OpenAI, Anthropic, Cohere, Mistral, etc), and, of course, NVIDIA.
If and when any part of this cycle starts to slow, it's going to have ramifications across the board. And really, the entire stock market.
For investors who back fast-growing start-ups, there is little downside to being wrong about the next big thing, but there is enormous upside in being right. A.I.’s potential has generated deafening hype, with prominent investors and executives predicting that the market for A.I. will be bigger than the markets for the smartphone, the personal computer, social media and the internet.
As I wrote the other day:
Unless your mentality – as is the mentality of many VCs – is that the worst thing in the world isn't to lose money, it's to not be in the best deals. And historically in the profession, this is the correct mentality to have. Removing the human elements from the equation, the fact is that you can only lose 1x your money if you pick the "wrong" deal. But if you pick the "right" one, you can garner a 100x, or 1,000x, or even possibly a 10,000x return. And it's actually potentially more than that because if you were able to pick the "right" deal, you could and would want to "double down" on such a deal, compounding your results.
So with that in mind, you would do basically every single deal if you could. Capacity is an issue, but at a high level, this is why some of the largest firms keep scaling (well that and the fee structure). Again, take any sort of emotion out of the equation: losing the initial capital is meaningless as long as you get one of those 1,000x or 10,000x returns. It's the cost of doing business.
There's a world in which this works – and to be clear, it will work for a handful of companies – but we don't currently live in that world. Again, there are several externalities that will lower the boom on the vast majority of these startups. To be clear, I don't want that to be the case – and actually, I'm heavily incentivized for that not to be the case! – but that's how I see this playing out.
AI – the blanket term – is the real deal. It is and will transform many things. But much of the investment in the space right now makes no sense on the case-by-case basis. Even though we'll all likely benefit and faster from these investments.
M.G. Siegler
M.G. Siegler
M.G. Siegler