The Rally That Left Skeptics Behind
It is worth remembering how quickly market sentiment can reverse. As recently as late March, the prevailing narrative held that the artificial intelligence trade had gotten ahead of itself. Commentators were calling for the end of the cycle, warning that valuations had outrun reality. Then the market did what it so often does to those who try to time it: from April through the following months, roughly 35% of all measurable inflows poured into the technology sector. It was a near-complete risk-on rally, and it left a great many cautious investors stranded on the sidelines.
That said, not every technology stock participated equally. The software names and various other corners of the sector lagged. The real engine of the move was, and continues to be, the data center theme. This is the part of the AI story that keeps gathering steam, and it is where the durable opportunity lies.
The Core Principle: Own Gushing Earnings, and Watch the Flows
The investment philosophy underpinning this thesis is simple and disciplined. We become investors in the first place so that we can share in a company's future profits. That means earnings come first. The companies worth owning are the ones that consistently beat expectations and then raise their guidance, signaling that the business is accelerating rather than merely holding steady. The word "acceleration" in a management commentary is a genuine reason for excitement, because it tells you the trajectory is bending upward rather than flattening.
The second pillar is verification through capital flows. It is not enough for a company to report strong numbers; you want to see real money moving into the stock and staying there. A name that attracts persistent inflows, with few or no outflows, is being validated by the broader market in real time. When strong earnings and relentless inflows line up, you have a high-conviction setup.
Three Companies Positioned for the Theme
Broadcom
The first name that stands out is Broadcom. The appeal here is the company's pattern of beating, raising, and then offering exceptional forward guidance. Analyst sales estimates for 2026 sit at roughly $103.6 billion, and projecting out two years, that figure roughly doubles to about $203 billion. These are precisely the kinds of growth curves an investor wants exposure to. On top of that, an enormous volume of capital is rushing into the stock, making it one of the highest-rated names in the data available. It is an excellent vehicle for expressing the data center thesis.
AMD
AMD's story is, once again, an earnings story. The company delivered one of the strongest guides of the entire earnings season, projecting revenue of about $11.2 billion at the midpoint, well past the roughly $10.5 billion analysts had penciled in. Management explicitly cited acceleration in both CPU and AI demand. Looking specifically at the data center segment, sales are estimated near $30.3 billion in 2026 and ramp to roughly $75 billion by 2028 — a multi-year trajectory rather than a single good quarter. Notably, recent earnings revisions pushed that out-year figure from around $65 billion to $75 billion, the kind of hockey-stick revision that reveals analysts running behind the actual pace of the business. The stock has more than doubled in a little over a month, and money flows into it have been a one-way train higher. The long-term case for AMD is compelling.
Corning
The third name moves the focus to optics. Photonics has become one of the most coveted themes in the market, and Corning sits squarely within it. The company announced a substantial multi-year deal with Nvidia — exactly the kind of marquee partnership you want to see from a great company in the data center space. In its optical communications segment, sales estimates for 2026 stand at about $8.35 billion, and following the Nvidia agreement, the two-year outlook stepped up to roughly $13.36 billion. That stair-step higher in estimates is the pattern to look for. What makes Corning particularly distinctive is its flow profile: over the past year of inflows, it has not registered a single outflow, something very few companies can claim, alongside multiple outlier signals in the data.
Nvidia and the Catch-Up Trade
It would be a mistake to read any of this as a knock on Nvidia. By common agreement it remains the standout stock — best of breed across the board, with a likely beat-and-raise that will probably stun the crowd and reinforce the conviction that AI is here to stay. There is genuinely little negative to say about the company.
The reason to favor the other names is not weakness in Nvidia but the dynamics of a broad catch-up trade now sweeping the market. For a long stretch, Nvidia was the crown jewel of nearly every portfolio. Now the "benchwarmer" stocks — names like Intel and AMD that investors had largely set aside — are getting the love. For portfolio managers, adding incremental risk through these older, somewhat forgotten companies is an easier and increasingly attractive way to express the theme.
Why This Is Still Early Innings
The buildout of data centers has years left to run, which means Broadcom, AMD, Corning, and many others form a durable base for the AI cycle to continue. The pattern of bottlenecks illustrates the point. A year ago, talk of a memory bottleneck struck many as far-fetched; today it is visibly playing out, and the constraint is now spreading into photonics and other areas. As data centers continue to be built across the country and around the world, more bottlenecks will surface — and each bottleneck is precisely where the next big opportunity is created.
This is also why the old mental model of investing is increasingly obsolete. For decades, investors thought in terms of sectors. The market no longer moves that way. It moves thematically, along frontier lines. Recognizing that shift is how an investor stays ahead of the curve, and it reframes the AI buildout as a multi-year runway rather than a stretched late-cycle trade.
Geopolitics as a Sideshow to the Earnings Story
High-profile meetings between technology executives and counterparts in China look constructive optically. As long as the relevant parties come to the table with their concerns and a willingness to work together, the eventual news flow is more likely to be positive than not once everyone is on board. But geopolitics, in this framing, is largely a sideshow. The dominant question is fundamentally an earnings question: where is the money going? It is flowing into all of these pockets of the data center economy, and the leaders across the world will ultimately find a way to navigate the politics, because the underlying technological revolution is simply too large to derail — a shift of a magnitude not seen in a long time. It is genuinely difficult to find a negative when the analysis is conducted on a multi-year horizon.
What to Listen For Next
Heading into Nvidia's report, two themes deserve close attention. The first is photonics. There are several smaller-cap companies not yet in the spotlight that the market is watching for an "anointing" moment — the point at which they are recognized as genuine photonics plays. That recognition tends to draw intense investor focus, because photonics is among the hottest themes in the market and one of the most frequently discussed by sophisticated investors.
The second theme is memory. Many investors worry about margins and, more specifically, about how much supply a company has actually captured. Commentary on the memory situation will be a key tell. Across just about every earnings report in the past year-plus, the story has been the same: better visibility, higher numbers, and more partnerships with each cycle. Given that consistent pattern of upward revision, there is a reasonable case that even today's elevated estimates may still prove to be too low.