Everyone's buying AI chips. The real shortage is electricity.

If you have invested for any length of time, you have almost certainly been right about something and still lost money on it. You saw the theme early, you bought in, and then the trade re-rated so violently that you were shaken out before it ever paid off — or you held a name that the crowd had already priced for perfection. The story was correct. The position was not. That gap between seeing the trade and surviving it is the part nobody warns you about, and it is the whole reason the most interesting AI exposure today is not the one everyone is already crowded into.

Start with what almost no one disputes. Ask most investors how to play artificial intelligence and you will hear the same short list of chipmakers and hyperscalers — the names already in the headlines, already in the funds, already priced as if the future is a foregone conclusion. That is the obvious trade. And the obvious trade is rarely where the next dollar of edge is hiding.

Here is the part that gets less attention. Training and running large AI models is extraordinarily energy-hungry. A single large data-center campus can draw as much power as a small city, and the build-out is accelerating faster than the systems that feed it. The scarce resource in the AI era is increasingly not compute — it is power, and the grid that delivers it.

The asymmetry nobody talks about

You can stand up chip capacity far faster than you can stand up the power to run it. A new fabrication line is measured in a couple of years. New generation and the high-voltage transmission to connect it — plus the permitting, environmental review, and interconnection queues that come with it — is routinely measured in five to ten. By various industry estimates, data-center electricity demand could rise substantially over the next decade; the open question isn't whether the chips get built, it's whether the grid can keep up.

The biggest buyers of compute already see it. Tech giants have begun signing long-term power agreements, including deals for nuclear capacity, and siting new facilities next to existing generation — moves that only make sense if energy, not silicon, is the binding constraint.

The trade hiding in plain sight

Follow that logic and the interesting exposure isn't only the household-name chip stocks everyone already owns. It's the unglamorous layer underneath: power producers, grid and transmission, electrical equipment, cooling, and the infrastructure that has to be built before a single new model can be served. It's a classic second-order trade — obvious in hindsight, overlooked in real time because it doesn't fit the headline.

The crowd buys the story everyone already knows. The edge is almost always one layer down — in what they overlook.

Which brings the conversation back to where it started. Identifying a theme is one thing; trading it well is another. Crowded narratives re-rate violently in both directions, and being early can look identical to being wrong for months at a time. You can be exactly right that power is the constraint, buy the right basket of producers and equipment, and still get knocked out of the position on the first 30% drawdown the narrative throws at you — because the hard part was never the analysis. It was holding the line when the screen turned against you.

Knowing where the puck is going is one skill. Sizing, timing, and the discipline to act on the move instead of the mood are another entirely. Which raises a more useful question than "what should I buy?" — namely, whether the buying and selling decisions can be removed from human judgment at the exact moments human judgment tends to fail.

What removing the decision actually looks like

That last question is the one Vector Capital was built to answer. The gap this article describes — that you can be right about the trade and still lose money executing it — is not a gap Vector tries to coach you through. It removes the moment of discretion entirely. Vector Capital is a systematic, market-neutral approach: a fixed rule set, defined in advance, that reacts to what prices are actually doing and is executed without a human override at the point a crowded narrative whipsaws. When the theme re-rates and most investors are deciding with their gut, the rules have already decided without one.

Because the rules respond to price movement rather than the story, the approach does not need to be right about which AI name wins, or whether the power thesis plays out on schedule. It trades the movement the theme creates — long or short, across stocks and futures — so it can act whether the crowded trade keeps running or violently re-rates. That distinction shows up most clearly in the years that punish conviction buyers: an environment full of sharp reversals is exactly the kind of condition the rules are written to act on, in either direction. The live figures, and the matching drawdowns, are shown below — and they are worth reading with the same skepticism you would bring to any track record.

How it actually makes money — whether the market goes up or down

Most people know only one way to make money in markets — the way they were taught. You buy something, a stock or a fund, and you wait, hoping it is worth more later than you paid. When it rises, you sell, and the gap is your profit. Buy, wait, sell higher. It works — but look at the catch buried inside it: you only win if the price goes up. If it falls, or sits flat for years, your money sits there with it.

There is a second way to make money, and most people never use it: you can profit when a price goes down. You take a position that pays off when something falls instead of rises — that is called going "short." Owning the normal way is going "long." Do both, and you can win whichever way the market moves, not just one.

"Market-neutral" means holding both kinds of position at once — some that pay when prices rise, some that pay when they fall — balanced so the market's overall direction barely matters to you. Picture a shop that sells both sunscreen and umbrellas. It does not need to guess tomorrow's weather; it makes money either way, as long as people keep coming in. A market-neutral system is the same. It does not need the market to go up. It needs the market to move — and aims to be on the right side of those moves, in both directions at once.

That is why returns like the ones below are even possible. An ordinary portfolio waits for one big upward move and prays nothing knocks it down first. A market-neutral system takes its profit from the movement itself — and movement is exactly what frightens everyone else. It is why a chaotic, fearful year like 2025 was the system's strongest, not its worst: more fear meant more movement, and more movement is more to trade.

The guarantee almost nothing else in finance will make

Now the part almost no one else in finance will put in writing. Vector backs the system with a 12-month satisfaction guarantee: if you are not satisfied with your first year, you get your money back.

Now think about everything else sold to people building wealth in their 40s and 50s. A bond locks your money up for years and hands you a fixed coupon — no refunds. A whole-life policy can take a decade just to break even, and surrenders at a loss if you leave early. An annuity charges you to get your own money back slowly. None of them — not one — gives you a year to decide whether it actually worked and then returns your money if it didn't. That simply isn't how financial products are built. The house does not hand the chips back.

A guarantee like that only gets offered by someone who has watched the system work across enough conditions — calm markets, crashes, melt-ups — to stand behind it with their own revenue on the line. It takes the risk off your side of the table and puts it on theirs. That is a very different proposition from being shown a number and asked to trust it.