Adobe and Marvell: Quality AI Plays Amid the Capex Re-rating

The investment thesis for AI is maturing. The initial, broad-based momentum play across the entire sector is giving way to a more selective hunt for quality and productivity beneficiaries. This shift is the core portfolio construction challenge today. The market is demanding proof of returns on massive AI capital expenditure, and it is rotating away from companies where that link is unclear or where the funding is straining balance sheets.

This bifurcation is already visible. While consensus estimates for 2026 capital spending by AI hyperscalers have climbed to $527 billion, investors have shown they are not willing to reward all big spenders equally. The rotation is clear: away from AI infrastructure companies where operating earnings growth is under pressure and capex is being funded via debt, and toward platform operators and productivity enhancers that demonstrate a clearer path to revenue conversion. The average stock price correlation among large public AI hyperscalers has collapsed from 80% to just 20% since June, a stark signal of dispersion and selective conviction.

This selective pressure is reflected in the broader market's technical picture. The S&P 500, a high-beta index heavily weighted toward these AI leaders, is flashing a Strong Sell signal based on moving averages and other indicators. The index's 14-day RSI of 23.341 suggests it is oversold, but the dominant trend remains down. This technical setup is a function of its composition; the sell-off in AI infrastructure stocks is dragging the entire benchmark lower, even as some quality names hold up.

The bottom line for portfolio construction is that the era of buying the AI story as a monolithic theme is over. The market's demand for proof of returns on AI capex is now the primary filter. For institutional investors, this means moving beyond simple sector bets. The opportunity lies in identifying the companies within the AI ecosystem that are not just spending, but converting that spending into sustainable earnings growth and cash flow. The next phase of the trade is about quality, not just capital intensity.

Identifying the Next Phase: Platform and Productivity Plays

The next leg of the AI trade is taking shape. As Goldman Sachs Research notes, the focus is shifting decisively from the capital-intensive infrastructure build-out to AI platform stocks and productivity beneficiaries. This is a fundamental re-rating. The market is no longer rewarding companies for spending; it is rewarding those that can demonstrate a clear, profitable link between that spending and future earnings. The divergence in stock performance among hyperscalers is the clearest signal of this new filter in action.

Within this new thesis, companies with durable financial strength and pricing power are emerging as defensive quality plays. Adobe exemplifies this profile. Despite the pervasive AI disruption narrative, the company's growth metrics have held up remarkably well. It posted revenue growth of 11% in constant currency last quarter, with strong subscription growth and robust operating income. Its massive cash flows have allowed for aggressive share repurchases, a capital return strategy that directly supports earnings per share. In a rotation away from debt-funded capex, Adobe's model-a high-margin, recurring revenue business with pricing power-becomes a relative safe haven. It represents a conviction buy for investors seeking to tilt toward quality within the broader tech sector.

This quality thesis is underpinned by a massive structural tailwind. The major U.S. hyperscalers are set to spend over $700 billion on capital expenditures in 2026. This isn't just a number; it's a multi-year demand engine for select component suppliers. While the market has bid up many semis, the opportunity lies in identifying the high-quality, niche players that are essential to this build-out but may be overlooked. Marvell Technology, for instance, is a key supplier of networking chips and custom AI accelerators for these giants. Its business model, which includes long-term contracts and a focus on cost-effective solutions, positions it to capture a portion of this spending. The key for portfolio construction is to identify these productivity beneficiaries-companies that are not just selling to the hyperscalers, but are embedded in the value chain that converts their massive capex into operational efficiency.

The bottom line is a move from pure capex exposure to earnings conversion. The institutional playbook is clear: overweight companies that demonstrate pricing power and cash flow generation, like Adobe, while selectively underweighting those where capex is straining fundamentals. At the same time, allocate to the high-quality component suppliers that are structurally positioned to benefit from the $700 billion+ spending wave. This is the next phase: a rotation toward quality, productivity, and tangible returns on the AI investment.

Valuation and Risk: The Quality Factor in a Crowded Theme

The AI trade has entered a phase where valuation and risk are paramount. After years of momentum, many AI-related stocks have pulled back, leaving a number trading at what appear to be attractive levels. This creates a classic institutional dilemma: a potential opportunity against a backdrop of heightened execution risk. The key risk is that the industry's massive spending does not translate into proportional revenue growth, a concern that has already driven the rotation away from less selective names.

The market's clear preference for quality is the most important risk premium signal. Investors are no longer rewarding capital intensity alone; they are demanding proof of returns. This is evident in the selective rotation away from AI infrastructure companies where operating earnings growth is under pressure and capex is being funded via debt. The divergence in stock performance among hyperscalers is the clearest manifestation of this filter in action. For portfolio construction, this preference for quality is not a suggestion-it is the new benchmark for risk-adjusted returns. It tilts the allocation toward companies with durable financial strength, pricing power, and a clear path to converting investment into earnings.

Marvell Technology presents a compelling case study in this new regime. The company is a key supplier of networking chips and custom AI accelerators for the major U.S. hyperscalers, positioning it directly in the path of the projected over $700 billion on capital expenditures in 2026. Yet its stock trades at an "incredible value" despite this excellent outlook. This disconnect between a strong structural tailwind and a depressed valuation represents a potential conviction buy on the quality factor. The risk here is not the macro spending trend, but the execution of Marvell's specific contracts and its ability to maintain its position in a competitive landscape. Management's recent guidance, which expects its custom silicon revenue to double year-over-year, provides a tangible near-term anchor.

The bottom line is that the quality factor is now the primary lens for assessing risk in AI. The valuation disconnects that have emerged after the pullback offer a chance to build positions in high-quality, productivity-enhancing businesses. However, the institutional playbook requires discipline: overweight companies like Marvell that are embedded in the value chain and demonstrate financial resilience, while underweighting those where rapid spending is straining fundamentals. In a crowded theme, quality is the ultimate differentiator for sustainable returns.

Catalysts and Watchpoints for Portfolio Managers

The selective rotation thesis now faces its first major test. The coming weeks will provide critical data on whether the market's focus on quality and productivity is a durable shift or a temporary re-rating. Portfolio managers must watch three key catalysts to guide future capital allocation.

First, the earnings updates from Micron and Accenture next week will offer a direct read on the AI-disruption narrative for software and infrastructure. Micron, a critical supplier of memory for AI data centers, will signal the health of the underlying hardware build-out. Accenture, as a major systems integrator, will reveal how enterprises are actually deploying AI budgets. Their results will test the divergence already visible in the hyperscaler group. If these companies show robust demand and pricing power, it will validate the thesis that the market is correctly rotating toward tangible productivity beneficiaries. A miss, however, could reignite fears that the spending wave is not translating into near-term revenue, pressuring the entire quality cohort.

Second, the key watchpoint is whether this focus on quality becomes a sustained sector rotation. The recent pullback in AI stocks has created valuation disconnects, but the market's reaction to earnings like Adobe's shows the narrative risk remains potent. Despite reporting revenue growth of 11% in constant currency and aggressive share buybacks, Adobe's stock sold off on AI disruption fears. This suggests that for software names, the quality factor may not be enough if the long-term competitive threat is perceived as existential. The rotation will only be structural if the market consistently rewards companies that demonstrate not just financial strength, but a clear, defensible moat in the AI era. Managers should monitor the performance of high-quality names versus those with strong fundamentals but higher narrative risk.

Third, the divergence between consensus capex estimates and actual spending must be monitored. Analysts have consistently underestimated AI spending, with the consensus for 2026 capex now at $527 billion. A continued gap between estimates and reality would signal a powerful, sustained tailwind for select suppliers. This is the structural driver for companies like Marvell. However, if actual spending begins to lag estimates, it could trigger a reassessment of the entire productivity-beneficiary thesis. The watchpoint is not just the level of spending, but the quality of the demand-whether it is being converted into profitable revenue at the customer level.

The bottom line is that portfolio construction in this phase requires a disciplined, event-driven approach. The earnings from Micron and Accenture will be the first major data point. The market's reaction to those results, and to subsequent earnings from other quality names, will determine if the rotation is deepening or stalling. For now, the quality factor remains the primary filter, but it must be paired with a close watch on the real-world conversion of the massive capex wave into earnings.