The Dominance of AI Infrastructure
AI infrastructure has become the unmistakable center of gravity in today's equity markets. The strength evident in semiconductor names and across the broader ecosystem of data center capex beneficiaries is remarkable, with leadership extending from chip manufacturers to industrial suppliers. Even when a bellwether like TSMC delivers a relatively tame monthly number — a roughly 1.5% dip on its latest report — the broader sector continues to rage higher, suggesting that single data points cannot dampen the structural momentum.
The memory segment has been particularly extraordinary. Micron, for instance, has surged about 14% in a single session and roughly 32% in a week, marking its best stretch since 2008. These are the kinds of moves that imply not just enthusiasm, but a genuine recalibration of how the market values exposure to AI compute demand.
Microsoft: An Underappreciated Position
Within the megacap universe, Microsoft stands out as attractively valued at around 22 times earnings, a multiple that reflects the broader compression seen across software in recent months. The fear weighing on the stock — and on software more generally — centers on seat-based business models. Investors worry that AI tools could displace or stall the growth of products like Microsoft Office, that Copilot may not be compelling enough, and that the existing revenue model is structurally vulnerable.
That fear has consumed so much investor attention that the strength of Azure has been overlooked. Azure revenue growth is accelerating and could hit or even exceed 40% next quarter. At the same time, Microsoft is integrating AI functionality across its software products, drawing on models from OpenAI, Anthropic, and a growing set of less expensive in-house models that improve existing offerings without piling on cost. Few software companies in the world possess the depth of customer ties Microsoft enjoys; that embeddedness with the global enterprise base is one of the most defensible moats anywhere in tech.
The Vertical Integration Divide
A meaningful narrative on Wall Street right now is that the megacap cohort is bifurcating along the axis of vertical integration. On one side sit companies developing their own silicon — Google with its custom chips and Amazon with Trainium — which gives them genuine cost advantages for running enormous AI workloads. Both have invested time, money, and engineering effort over more than a decade to build that advantage.
On the other side sit Microsoft, which still leans heavily on OpenAI for frontier model capability, and Meta, which lacks a comparable in-house chip program. Both are pushing aggressively to close the gap. Microsoft, in particular, is racing to develop internal silicon and to lower the unit economics of building and operating data centers. The trail position is real, but Microsoft compensates with the deepest enterprise distribution in the industry.
Capex Discipline and Spending Posture
The recent post-earnings selloff in Microsoft was driven not just by capex levels but by margin concerns. Yet a fair-minded look at the company's recent history suggests it has actually been quite disciplined — perhaps too disciplined. Roughly eighteen months to two years ago, Microsoft paused some of its data center build, worried about overbuilding. Within six months, it regretted the pause and scrambled to restart the cadence.
The balance sheet supports continued aggression. Net cash slightly exceeds debt, the dividend is healthy, buybacks have continued, and the company is generating free cash flow even amid heavy capital expenditure. The likely path forward over the next twelve months is more aggression from Microsoft, not less, mirroring the build-out posture of Amazon and Google.
Nvidia's Quiet Outperformance Narrative
Nvidia has been notching new all-time highs, yet a counterintuitive truth lurks beneath the headline: it has actually been one of the worst-performing semiconductor stocks over the past couple of months, as memory names and higher-beta, more momentum-oriented chip stocks have led the cohort. Within semiconductors, Nvidia has settled into the role of the steady stalwart.
Demand for the company's products continues to grow rapidly — it is selling everything it can produce, and that dynamic shows no sign of breaking. The valuation remains attractive, and a significant share of the upside in this AI cycle is likely still to be captured by Nvidia. There is room for continued earnings beats and further appreciation through the balance of the year. The recent Corning-related announcement also matters as a broader signal: it ticks the boxes around U.S. job creation, reindustrialization, and supply chain vertical integration that increasingly shape how investors evaluate the ecosystem.
The Industrial Beneficiary: United Rentals
Beyond pure technology, the AI build-out is rippling into industrial names. United Rentals serves two distinct customer bases. The first is local non-residential construction — hospitals, schools, and the everyday backbone of regional building activity. The second is mega projects: massive battery manufacturing facilities, semiconductor fabrication plants, airports, and similar undertakings.
The mega project segment has been performing extremely well and continues to do so, lifted by the same reindustrialization themes powering broader markets. The local non-residential side, weak for the past year and a half, is finally inflecting. That turn matters because it is creating an emerging shortage in used non-residential heavy construction equipment, which in turn hands pricing power to United Rentals and its peers — pricing power they have not enjoyed in the last two years.
The combination of improving demand, improving pricing, an active cost-cutting effort lifting margins, a leading position in a growing industry, and disciplined capital allocation makes United Rentals a compelling through-the-cycle compounder. Even after a strong move on the back of recent earnings — the stock is up roughly 15% year-to-date and around 40% over the past 52 weeks — there is still a credible case for buying it at current levels.
Synthesis
The current market is rewarding both ends of the AI value chain: the chipmakers and memory producers feeding the data center boom, and the industrial firms providing the equipment that physically makes those facilities possible. In the middle sit the hyperscalers, whose differentiation increasingly hinges on whether they own their silicon. Microsoft, despite trailing on the chip front, retains a uniquely deep enterprise franchise and a balance sheet built for sustained aggression. Nvidia, paradoxically, looks underappreciated even at all-time highs because its semiconductor peers have outpaced it. And companies like United Rentals quietly capture a slice of the same secular wave through the unglamorous but durable mechanics of equipment rental. Taken together, the AI infrastructure theme is broader, deeper, and more diversified than the headlines about any single chip company suggest.