Wall Street loves an early leaderboard. Since the breakthrough of generative artificial intelligence, a substantial share of the analytical work has been about assigning roles: those who will supply the chips, those who will own the models, those who will control the data and those whose product will soon become useless.
The question is no longer whether there will be winners and losers. There will be, and the market believes that it has already identified them. However, be careful not to move too fast; the difficulty lies precisely in thinking that you can identify them when the transformation has barely begun.
Major technological breaks rarely have the courtesy to follow the timetable markets imagine. The internet is a useful precedent. In the early 2000s, it was reasonably easy to understand that digitization would weaken certain business models. It was much harder to determine which would be destroyed quickly, which would adapt, and which would keep thriving for ten or fifteen years.
The overall direction of change could be anticipated correctly, even as its distribution was completely misjudged. AI could reproduce this phenomenon on a far larger scale.
A certain disruption says nothing about the right price
Financial markets have a particular problem with technological breaks: they must now value consequences that will only become observable gradually.
When a new product threatens an industry, waiting five years to know precisely who will survive is obviously not an option. Prices must immediately reflect a probability of destruction. That encourages shortcuts. A sector is threatened, so its players must be discounted.
Sometimes the reasoning is perfectly justified. It becomes dangerous when it confuses exposure to a break with the disappearance of a business model.
A company can operate in a structurally declining industry and still be a profitable investment. That only seems contradictory when you forget the price.
Imagine a business destined to gradually lose customers. Its future growth is nonexistent and its terminal value mediocre. Yet if its assets are bought for a fraction of their economic value, if the company keeps generating cash for several years, and if management returns that capital rather than funding a costly attempt at reinvention, the resulting return can be remarkable.
The investor does not need to be wrong about the sector's decline. They can fully acknowledge the business will disappear and still believe the market is paying too much for that disappearance.
This distinction is particularly useful for analyzing companies threatened by AI today.
A company does not need a bright future for its shares to be undervalued. It simply needs a future that is less bad than what is already embedded in the price.
Buying a decline is still a matter of price
This discipline becomes especially important with AI because the technology invites two excesses. The first is to treat every exposed company as doomed. The second is to treat every AI-driven selloff as a buying opportunity.
Some presumed victims will be real victims.
A publisher whose product can be easily replicated, whose customers face little cost to switch providers, and whose cash flows depend on prices that have become unsustainable is not saved simply because its bond now offers a higher yield.
The picture changes for software that is deeply embedded in customers' operations. Replacing such a tool can mean migrating data, rebuilding processes, training employees, connecting new systems, and accepting months of operational risk. AI can reduce the vendor's future growth without making those switching costs disappear.
You have to set aside sweeping statements about disruption and return to far more mundane questions: How many customers will actually leave? How fast? How much cash will remain? What debt will have to be refinanced? What assets can be sold? What value will be left if the scenario deteriorates further?
And above all: how much is the market paying to bear these risks?
After the euphoria comes the sorting
One heavier unknown remains.
Artificial intelligence related spending has grown so large that it now contributes to the broader investment cycle. A deceleration would therefore have consequences that go beyond technology companies alone. Growth expectations, financing needs, and potentially the price of credit could all be affected.
Above all, the market will eventually demand accountability for the hundreds of billions invested.
The first years of a technological revolution easily tolerate promises. The years that follow become more ruthless. Companies will have to show that the spending actually improves productivity, lowers costs, or creates revenue large enough to compensate the capital deployed.
Some will succeed. Others will have mostly accumulated capacity whose economic profitability will prove mediocre.
Dispersion should then gradually replace the collective enthusiasm.
The internet precedent offers a useful signal. At the turn of the century, the market was right about the essentials: the internet was going to reshape the economy. That remarkable intuition prevented neither valuation mistakes, nor companies written off too early, nor those that were funded for far too long.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.