Remember the exact moment you first typed a question into ChatGPT and got a jaw-dropping answer back in seconds? For millions of us that happened almost exactly three years ago, on 30 November 2022. What felt like a clever toy that weekend has since rewired entire industries – and sent certain stock prices into the stratosphere.
I still catch myself grinning when I think how quickly everything moved. One minute we were marvelling at a chatbot that could write decent poetry; the next we were watching a chip company most people had never heard of become the most valuable company on Earth. That company, of course, is Nvidia – and its wild ride is the perfect symbol of everything the past three years have been about.
Three Years That Changed Investing Forever
Few product launches in history have had such an immediate and measurable impact on public markets. Within months of ChatGPT going viral, the phrase “generative AI” was on every earnings call, every investor letter, and every CNBC segment. Money poured into anything even vaguely related to the technology.
The numbers are frankly ridiculous. Nvidia’s revenue has multiplied roughly ten times since late 2022. Its market cap sailed past $1 trillion, then $2 trillion, then $3 trillion, and at one point this autumn kissed $4 trillion – briefly even nudging $5 trillion before reality bit back. No company has ever grown that fast in absolute dollar terms.
And it’s not just Nvidia. Cloud giants, data-center landlords, even copper miners and electric-utility stocks have enjoyed what traders now call “the AI trade”. Mention you’re building a hyperscale training cluster and watch your share price jump overnight.
The One Company Everyone Is Watching
At the centre of this tornado sits a company most retail investors still can’t buy directly: OpenAI.
Every major announcement from the ChatGPT creator moves markets. When word leaked that Nvidia might commit up to $100 billion (yes, billion) in hardware purchases over time, Nvidia shares leapt almost 4 % in a single day. When Oracle revealed OpenAI could spend hundreds of billions on its cloud, Oracle stock rocketed 36 % – only to give much of it back when scepticism crept in.
“Public companies that have adopted and run with AI have been rewarded and those who have not have been punished.”
Victoria Scholar, Interactive Investor
But Here’s the Awkward Question Nobody Wanted to Ask… Until Now
All of this frenzy rests on a fairly simple assumption: that the companies spending eye-watering sums on AI infrastructure will eventually make even more eye-watering sums from selling AI products to the rest of us.
Three years in, the evidence is… mixed.
Recent research from major investment banks paints a sobering picture. Even under extremely optimistic scenarios – user base growing from roughly 10 % of the world’s adults today to 44 % by 2030, revenue exploding from around $12 billion this year to over $200 billion in 2030 – OpenAI is still projected to remain deep in the red for the entire decade.
One particularly stark forecast suggests cumulative cash shortfalls could top $200 billion between now and 2030, even after massive external funding rounds and hardware credits. The main villain? Compute costs that rise almost as fast as revenue.
- 2025 cloud compute bill: already in the tens of billions
- 2030 estimate: north of $200 billion per year for OpenAI alone
- Break-even on compute costs versus revenue: possibly not until the early 2030s
In plain English: the company widely seen as the leader may not turn an annual profit for another eight to ten years – if ever in its current form.
Are We Staring at Another Dot-Com Bubble?
I’ve lived through a few market manias, and the parallels are hard to ignore. Enormous capital expenditure today justified by vague promises of future productivity gains tomorrow. Valuations detached from current cash flow. A single dominant narrative drowning out caution.
That doesn’t mean the technology is fake or worthless – the internet in 1999 wasn’t fake either. But it does mean timing matters enormously, and many of today’s sky-high valuations are pricing in a flawless execution that history suggests is unlikely.
“Long-term stability will matter far more than rushing into the public markets.”
Dan Moczulski, eToro UK
The IPO Everyone Is Whispering About
Speaking of public markets, rumours continue to swirl that OpenAI itself could list as soon as the second half of 2026. Some reports suggest a valuation as high as $1 trillion – roughly double its most recent private round.
Whether that actually happens depends on many things, not least the company’s ability to convince investors that a path to sustainable profitability really exists. Advertising on the free tier, enterprise licensing, new consumer products – all are being explored. But none have yet moved the needle enough to quiet the sceptics.
What Should Investors Actually Do Right Now?
First, a reality check: the AI megatrend is almost certainly real and will reshape the global economy over the coming decades. The question is pace and winners.
- Stay exposed – but favour companies with actual earnings today (think the large cloud providers, Nvidia itself while margins remain astronomical, even some specialised software firms)
- Be brutally selective – pure-play infrastructure bets that lose billions annually are speculative in the extreme
- Watch the compute cost curve – any sign that training costs are falling faster than expected would be enormously bullish
- Keep powder dry – corrections in over-loved themes can be sharp and offer far better entry points
Personally, I’ve trimmed some of my most stretched AI-related holdings this autumn while adding to broader market exposure. The story is far from over, but three years in feels like a sensible moment to ask whether the market has sprinted too far ahead of reality.
ChatGPT’s third birthday is worth celebrating. It really did kick off something historic. But as any parent knows, year four is when the toddler starts running – and occasionally falling over.
Let’s hope the AI industry has good knees.
