I was at the Y Combinator Demo Day recently, talking with another investor in Silicon Valley about an early-stage startup that had just raised seed funding. Needless to say, it was an artificial intelligence company. As we discussed the financing round, I inadvertently remarked that the startup had been valued at “a few billion dollars.” I immediately corrected my slip of the tongue, since I meant “a few million dollars” instead. But what stayed with me was not simply that I had misspoken but the fact that the word “billion” had blurted out of my mouth. Then, I realized that somewhere along the AI run-up, the word “billion” had crept into my intuitive dictionary and cemented itself in my mind. We hear the word so frequently in the AI financing rounds that it doesn’t surprise us anymore.

A billion dollars is still an enormous amount of money. But psychologically, our unit of measurement has intuitively evolved. A decade ago, reaching a one billion valuation was such a rare accomplishment that Silicon Valley created a special word for it: unicorn. Today, for some AI companies , a billion-dollar valuation is the starting point.

Some companies are receiving enormous valuations before they have demonstrated the milestones traditionally expected of startups. Some even have small teams with no product out yet, let alone having any revenue. Instead, they have team branding and pedigree in the AI sector. Their founders may have built successful AI companies before. They may be professors at leading universities, or they may have spent years inside leading frontier AI labs . So, for a small number of founders, reputation itself has become an extraordinarily valuable asset. If a team contains researchers believed capable of building a breakthrough AI model, inference architecture, chip or agent, investors may be willing to assign substantial value before the company has demonstrated any semblance of a business model. In those cases, investors are effectively underwriting human capital before commercial proof.

The Reversal of Funding Sequence

The traditional startup journey in the pre-AI era followed the sequence of product and customers first before attaining a high valuation. But in parts of the AI sector, the sequence now begins with a large valuation followed by product and (hopefully) revenue. I feel nervous adding the word ‘hopefully’ because not all highly valued startups may reach the revenue level needed in time to justify and sustain the lofty valuation.

That does not necessarily mean that investors have become irrational. It means they believe the upside has become so large, given an infinite AI market size and the scarce exceptional technical talent, that waiting for traditional proof may mean investing too late. So, there is logic to that approach.

Lofty Valuations May Be Justified

People say Silicon Valley is living in the Wild, Wild West era . But it is important to look at the investors’ justifications behind these numbers.

Firstly, as mentioned above, the perceived markets are enormous. Investors don’t view AI as another software category. They increasingly see it as a technology that can replace labor in every sector it can touch, whether it is software engineering, financial services, recruiting , healthcare, education, advertising, manufacturing, mortgage or robotics . This expands the addressable market for venture-backed companies by an order of magnitude.

Secondly, some AI companies are scaling at speeds rarely seen in previous generations of software. If investors genuinely believe a company can go from almost no revenue to tens of millions of dollars remarkably quickly, conventional valuation frameworks can not be applied.

Thirdly, parts of AI are extremely capital-intensive . Building models, semiconductor systems, robotics platforms or inference infrastructure can require enormous amounts of compute and energy. For companies building a chip, securing fab manufacturing capacity alone can cost a couple hundred million dollars.

Finally, venture investing has always been governed by the power law . While investors ideally want every company to succeed, they know that a small number of companies can generate outsized returns. If a future valuation can reach a hundred billion dollars, paying a billion-dollar valuation for an early position is justified. The problem, of course, is that only a small number of companies will ultimately become extraordinary outcomes, but that might be fine for investors at a fund level.

Eventually, all valuations have to be justified by real business metrics. AI market may create enough economic value to justify lofty numbers that once appeared impossible. But AI may not justify every number attached to every company. Time will tell the answer. Until then, Silicon Valley will keep speaking the language of billions. Because in the age of AI, a billion is starting to sound like a cent.