A VC guide on how to find an AI bubble

The looming AI bubble could be about to burst. If you know where to look, there are signs of corrected telltails everywhere.

It may be too early to say whether AI is the “five industrial revolution” or a great way to lose money, but onlookers like Nnamdi Okike, the founding partner of 645 venture, use heuristics to distinguish between game-changing advancements in AI and floss and fools.

“One drawback to the size of the rounds that are occurring and how quickly these rounds are occurring is that investors may lack this idea of ​​the quality of their business models,” Okike told Business Insider.

The company will support startups in Series B from before seed with notable investments, including Meridian, an AI-enabled fintech platform for private equity trading. 645 led the $7 million seed round in June. Last year, it also led to SetPoint, a lending infrastructure company, $31 million in Series B funding.

Within the class of investors, the final conclusion of the AI ​​boom compares with the vibe of a crypto fraudster, so many feel that they've seen this all.

“It is very difficult in these early days to truly understand what a sustainable business model is with an attractive long-term margin.

Okike walked me through his investment standards and whether he was in the AI ​​bubble already tied to pop.

This Q&A has been edited for clarity and length.

Talk about the company's investment spirit, particularly how it invests in AI or not.

The latest deal I was involved with was with Arbor, software that effectively replaces operations consultants.

This is an example of an AI company that offers business intelligence. Historically, companies may have had to go to a high-priced consulting company to do specific business intelligence projects rather than using software in real time.

We've moved away from a very capital-intensive business. For example, we don't invest in a wide range of large language model companies.

We are looking at a seed round where you can invest and buy 10% stake in the business for $2-3 million. Even in Series A, we are considering rounds of $10 million to $12 million. There, they are valued from $40 million to $60 million in post money.

We are not investing in companies raising $2 billion Series A.

Can you explain in detail why some AI companies are seeing such high praise?

Companies skip multiple steps and increase rounds that are far more worthy of the size of companies that have historically had a truly established fit for a product market.

When you start to lose it, you need to prove so much after money, and you are often doing it without the necessary rarity and capital efficiency.

I'm a huge fan of quotes. “Need is the mother of invention. This forces trade-offs, forcing them to say, “You can only spend X on an engineer” or “You can only use it to go to the market.”

If you can't do that, it's often proof that one of your assumptions is wrong, your performance is bad, or that the market forces you to change. That's good in the entrepreneurial context.

When you leave that system, it ruins things where we are. This is a very large round that is much more consistent with what you'll see in the ultra-IPOs that appear before Series A.

This will likely lead businesses to raise rounds and will likely waste their money usage. Hidden variables are also the quality of revenue.

Is that a high margin revenue? Is that a defensible revenue? Whether it's legal AI, financial AI, or developer tools, you'll see a dramatically bulging round at every stage.

We write about special purpose investment means, SPVSand Spack, which contributes to the fear of the AI ​​bubble. Why do these types of investments show that?

If you check out blank checkspacks, or spacks that are raised without really clearly focusing on the type of company you are buying, I think investors are somewhat blindly investing and trusting owners to understand that. Generally speaking, the last kind of SPAC bust group proved that.

Let's go back to why SPACS exists. There are companies designed to buy attractive assets that are inefficient in the private market and invest in them, or if the open market is the right place for those companies to exist, there are companies designed to buy attractive assets.

Some of the examples of SPACS that were good investments in the final wave were DraftKings and HIMS. In the case of DraftKings, the SPAC structure allowed multiple businesses to merge, which was complementary.

If there is a disconnect between how private and public markets view business, they are wrong.

The public market is so bubbled up, my guess is that SPAC investors “OK, there is a lot of demand for AI companies, so national market investors have access to theoretical attractive AI assets, and they trade at premium.”

If you see AI companies trading on 50 or 100x revenues, or increase the rounds with that, what is the final game? If it is ultimately valued in the open market with 5-10x revenues, it will need to grow dramatically to overcome that valuation reduction.

In addition to investing in AI, what other metrics are there heading towards the AI ​​bubble?

There are some common signals for bubbles, and they are relatively popular in different hype cycles of technology. 1) a rapid increase in valuations, 2) an increase in frequency of new funding rounds, 3) what investors or other investors are trying to do in a larger fool theory, basics, 4) a rise in price based on basic indicators, price to revenue, revenue, etc.

Openai's valuation went from $300 billion in the final round to $500 billion in the tender offer. It is very rare for private companies to be valued at $100 million. Now more companies are adding ratings and increasing ratings within a few months.

Large companies are valued at a much larger percentage of the public market, which can affect the performance of these markets in the long run. Looking at the grand Seven, you can see that while they are not valued at a very high multiple of revenue, their capital expenditures rise dramatically.

It is estimated that $560 billion has been invested in CAPEX this year, about $35 billion in revenue generated by that spending. Openai is about $10 billion for this, which is essentially buying Microsoft Azure and Services. So we'll save a bit as Microsoft sells it at a cost. There is a major disconnect between CAPEX expenditures and the revenue generated by those expenditures.

Given that these companies are concentrated in large markets, changes in their spending can cause a chain reaction.

You can also see that certain public companies trade with very high revenues. One is Palantia. It trades more than 100 times the revenue, which is extremely extreme and rare. Palantir is a great company, but when you start looking at multiples of them they say, “Hmm, that reminds you of the previous bubble period.”

If your conclusion about the AI ​​bubble is correct, what will we start to see, and what will the big technology fallout be?

There are often revisions in the market cycle, and revisions usually focus on companies tightening their wallet strings and profitability and high return investments.

Ironically, there was a revision after the pandemic-driven 2021 tech in place. The focus is on reducing profitability and spending in both the private and open markets, and there was the idea that these unicorns should raise rounds, increase profitability and focus on efficiency, but then AI came into play, and many of them were dumped through the windows.

You really didn't have a complete fix. Some of them still rock themselves. I think the same thing could happen after this boom with AI. Do you think the question is how important it is and whether it is permanent? And at some point, are businesses really forced to tighten their wallet laces?

This means that CAPEX investments in epic Seven, perhaps small venture rounds, will be much less, and companies must be more efficient in how they train their models.

What's interesting is that I had this data point from DeepSeek. The market has responded very dramatically to this one-off company that can train models in just a fraction of what US LLM trains models.

The market seems to have forgotten that, and businesses continue to invest at the rates they were. It's a strange outlier event.