The AI Boom Has a Clock Problem (And It’s Your Biggest Opportunity)
Listen to the audio version of this article (generated by AI).
Let’s go back in time, to the Atlantic in 1761, as a ship groans westward through swells that have swallowed better vessels than this one. It’s carrying cargo that cannot be insured, cannot be replaced, cannot even be fully explained to the men guarding it.
It is a watch.
Not gold. Not flush with jewels. Just a simple watch, ticking away in a box belowdecks.
It doesn’t make the ship any faster. It doesn’t calm the violent Atlantic waves. But it may answer the question that has haunted seafarers since the first hull left sight of shore…
Where, exactly, are we?
For generations, that question had no reliable answer. Latitude, sailors could find by the stars. Longitude — a measure of position running east to west — could only be calculated if sailors carried an accurate record of time from their home port across an ocean whose rolling and humidity constantly threatened to throw every timekeeping mechanism off course.
Enter John Harrison.
Harrison was not a scientist by training, but a carpenter, self-taught in the language of gears and escapements. He had spent most of his adult life chasing a single obsession: a watch that could survive the sea.
His fourth attempt, H4, was the watch aboard that ship. His son William held it in trust, checking it daily, trusting it over the instincts of men who had sailed these waters for decades.
And when H4 predicted the crew would spot Madeira before the sailors believed it possible… It was right.
The island appeared exactly when the little watch said it would.
A small machine, no larger than a dinner plate, had just quietly solved one of the most expensive, most deadly problems in the history of navigation.
There’s a lesson buried in that story: an enterprise of staggering scale can hinge on a component few think to look for.
I’d argue the same principle is quietly reshaping the AI economy today.
Everyone is watching the chips, or the data centers, or the spending commitments stacking into the hundreds of billions.
Almost no one is watching the clock.
Yet precision timing has become one of the unglamorous necessities of the AI buildout. It’s the synchronization that lets massive, expensive systems operate as one coherent machine instead of a thousand disconnected parts. In the latest episode of Being Exponential with Luke Lango, we examine a company doing exactly that.
It’s a timely opportunity, if you’ll forgive the pun, because the broader market right now feels like two different economies wearing the same face.
Borrowing costs still bite. Consumers are stretched thin. Half the stock market looks nothing like the half that keeps making headlines.
So which is it? Should investors wait for conditions to improve, or are they hunting for growth in all the wrong corners?
My take: the AI earnings engine is accelerating even as the rest of the economy strains under its own weight. Understanding that divergence is how you find proven winners.
So let’s follow the thread from the market’s current tug-of-war, to an overlooked component supplier keeping time for billion-dollar machines, to a financing headline that looks alarming right up until you check the math.
Earnings Are Pulling Harder
The market framework I keep coming back to is simple: higher interest rates versus higher earnings.
Higher rates make borrowing more expensive and reduce what investors are willing to pay today for profits expected years from now. That creates a headwind for growth stocks.
Rising earnings pull the other way.
In the episode, we discuss encouraging signs from across AI infrastructure: Marvell’s stronger outlook, AMD CEO Lisa Su’s comments about demand and supply constraints, and similar demand commentary from Nokia.
Those signals matter because they help connect enthusiasm about AI with the businesses supplying it.
A bigger addressable market alone doesn’t pay the bills. Orders, shipments and profitable revenue do. The more spending converts into those results, the stronger the earnings side becomes.
That’s why stocks can advance before the interest-rate problem disappears.
Suppose a company earns $5 a share and trades at 30 times earnings. That’s a $150 stock. If earnings rise to $7 while its valuation falls to 25 times earnings, the stock reaches $175.
Investors paid a lower multiple. The share price still rose.
That’s an illustration, not a price target. But it explains why waiting for an all-clear on rates can mean missing an earnings-driven advance.
Follow the Spending Beyond the Obvious Names
This also explains why a rising index can coexist with a lot of struggling stocks.
A consumer business depends on households having the money and confidence to spend. An AI infrastructure supplier may depend on technology companies expanding computing capacity.
Those are different customers with different budgets.
That’s the AI Bifurcation we keep discussing. You can have weakness in one part of the economy and powerful investment growth in another.
For investors, the useful question becomes: Where does the AI spending actually go?
Some goes to computing providers such as Nebius (NBIS). Some goes to networking and semiconductor suppliers. Some goes toward the electricity required to run everything, which is why we also discuss Constellation Energy in the episode.
These businesses offer different forms of exposure, with different financing needs, competitive pressures and valuations. Being farther from the computing layer doesn’t automatically make a stock safer.
But it does expand the opportunity set.
You don’t have to predict which chatbot wins every customer to investigate the businesses supplying infrastructure that many competing platforms need.
The Tiny Clocks Behind the Giant Machines
Which brings us back to Harrison’s watch, and to SiTime (SITM).
SiTime makes precision timing products. Think of them as part of the heartbeat of electronic systems.
Inside an AI data center, processors, networking equipment and communications links must coordinate activity. Timing errors can create delays and waste expensive computing capacity.
The financial logic is straightforward: If a component helps costly equipment do more useful work, its importance can far exceed its size.
Traditional timing devices often rely on quartz. SiTime builds products using silicon MEMS — microelectromechanical systems — with advantages in demanding conditions, including temperature changes and vibration.
In May, the company introduced a timing product specifically designed to improve synchronization and GPU utilization in AI data centers. That is a concrete example of timing technology being developed around AI’s operating demands.
I see two potential growth drivers here.
First, the infrastructure buildout creates more systems that need timing components.
Second, increasingly demanding systems can create opportunities to upgrade those components.
That combination is why I’m enthusiastic about SiTime. More infrastructure can expand demand, while better performance requirements can create room for more advanced products.
It doesn’t make the stock a buy at any price. The investment case still depends on execution, adoption and valuation.
But it illustrates exactly the kind of overlooked business I want investors to understand.
Track the Customer, Not Just the Construction
Of course, infrastructure enthusiasm needs an economic foundation.
That’s why we follow revenue growth at OpenAI and Anthropic so closely. Their ability to turn AI usage into paying business is an important signal about the demand supporting this buildout.
One measure discussed in the episode is annualized revenue run rate, or ARR.
Think of it as a snapshot of the current revenue pace extended over a year. It is not necessarily the amount already earned over the previous 12 months, guaranteed future revenue or profit. Definitions also deserve scrutiny.
The useful signal is the trajectory: Is paid demand continuing to expand fast enough to support the investment happening around it?
For diversified companies, that demand can be harder to isolate. Cloud growth can provide clues, but cloud revenue includes more than AI.
No single number settles the debate. I want to see customer monetization, supplier revenue and infrastructure commitments reinforcing one another.
A Scary Headline Still Needs a Denominator
The episode also tackles a subscriber’s concern about SoftBank borrowing against its Arm (ARM) shares.
Borrowing against a volatile asset creates risk. But the loan amount alone doesn’t tell you how much.
Using the figures discussed in the episode — 769 million pledged shares, an assumed $300 share price and a $25 billion loan — the collateral would be worth approximately $230.7 billion.
That puts the loan at about 10.8% of collateral value.
Under those assumptions, Arm would have to fall roughly 57% for the ratio to reach 25%, assuming the debt and pledged share count stayed unchanged.
That 25% figure is an illustration, not a verified margin-call trigger. Actual lending terms matter, and a low starting ratio doesn’t eliminate refinancing or forced-selling risk.
Still, the exercise helps separate a frightening headline from the financing question underneath it.
Before deciding how worried to be, do the math.
What I’m Watching Next
I’m constructive on the outlook for November and December because I think earnings can keep strengthening while some of the pressure from rates eases.
That is a forecast. The timing of geopolitical relief and changes in borrowing costs remains uncertain.
What would challenge my view? Slower AI monetization, weaker supplier outlooks or meaningful cuts to infrastructure spending. Those would matter more than a reassuring headline about the calendar.
For now, I’m focused on businesses turning AI demand into revenue — and solving problems that become more important as the buildout grows.
In the full episode of Being Exponential, Brooke and I work through the market outlook, the SiTime thesis and the Arm financing question in more detail.
Watch the full conversation to see how those pieces fit together, and leave us a question for a future episode.
Harrison’s watch helped sailors locate their ships. For investors, the exercise is locating the indispensable parts of this enormous buildout.
Some of them are very small indeed.
P.S. My colleague Jonathan Rose has an ambitious goal: Show folks how they could turn $10,000 into $100,000 using his trading strategy. On Wednesday, October 14, at 8 p.m. Eastern, he’ll explain his approach during the free $10K to $100K Challenge. Jonathan brings decades of professional trading experience — and a knack for making the lessons understandable. Sign up here to see how he finds opportunities, plans his trades, and decides what he’s willing to risk.
