July 24, 2026: The three major US tech giants ended the trading day with divergent performances. Microsoft closed at $381.70, up a modest 0.031% on the day. Meta finished at $595.19, down 1.80%. Amazon settled at $232.11, falling 0.66%. These mixed results mirror the market’s complex mood ahead of the upcoming earnings season.
Microsoft will release its fiscal Q4 2026 earnings after the market closes on July 29. Meta and Amazon will also announce their Q2 results on July 29, while SK Hynix is set to disclose its Q2 earnings the same day. This dense earnings window for four major companies will collectively address a central question that’s dominated 2026: When—and how—will the hundreds of billions invested in AI infrastructure by tech giants translate into verifiable revenue and profit?
Over the past 18 months, the dominant narrative in the AI sector has been "scale of investment"—who spends more, who has more data centers, who buys more GPUs. But this logic is shifting. The market is no longer satisfied with the speed of spending; it’s now focused on the efficiency of that spending.
Microsoft: Balancing Azure AI Growth and Capital Expenditure
Microsoft’s earnings logic boils down to a simple arithmetic question: Can Azure’s growth offset its ballooning capital expenditures?
Analysts generally expect Microsoft’s Q4 revenue to reach about $87.4 billion, up 14.3% year-over-year, with EPS forecast at $4.21, a 15.3% increase. Azure, the company’s growth engine, is guided to grow 39% to 40% at constant currency. Goldman Sachs projects Azure’s Q4 constant currency growth at 40% to 41%; Deutsche Bank expects similar numbers.
The real wild card for Microsoft’s stock price is the guidance on future capital expenditures. Last quarter’s $31.9 billion spend already raised eyebrows. For all of 2026, Microsoft’s AI capex is estimated at around $190 billion. The core concern: Does this scale of investment signal "unrestrained spending"?
On the revenue side, Microsoft’s AI business is running at an annualized rate of $37 billion, with over 20 million paid Copilot seats. Since July 1, Microsoft 365 Business has integrated Copilot as a standard feature, marking a shift from add-on to default—potentially pushing paid Copilot seats past 25 million by year-end. Morgan Stanley believes Azure growth will inflect in the second half, as new capacity converts to revenue and Copilot’s enterprise monetization potential remains underestimated.
However, the bearish case is equally clear. If capex guidance far exceeds the market’s expectation of about $220 billion for fiscal 2027, concerns about free cash flow pressure will intensify. Microsoft’s current P/E ratio is about 22.7x, well below its five-year average—meaning the market has already priced in some pessimism. But if earnings fail to dispel the "capital black hole" narrative, the scope for a valuation rebound will be limited.
Meta: Ad Revenue Keeps Flowing, But When Will AI Investment Pay Off?
Meta’s situation contrasts subtly with Microsoft. Its ad business remains robust, but AI spending is eating into margins.
Meta’s Q2 revenue guidance ranges from $58 billion to $61 billion, implying a midpoint growth rate above 25% year-over-year. In Q1, Meta’s family of apps generated $55 billion in ad revenue, up 33%. Ad impressions grew 19%, and average ad price rose 12%. Video consumption on Instagram and Facebook hit record highs.
Yet the scale of AI investment is becoming a significant financial burden. Meta raised its full-year 2026 capex guidance to $125 billion–$145 billion, up from the prior $115 billion–$135 billion range. Consensus expects Q2 EPS at $7.13, slightly below last year. In other words, strong ad revenue growth may only offset the margin erosion from AI spending, rather than drive net profit growth.
More notably, Meta still lacks a clear roadmap for AI commercialization. The company admits it doesn’t have a "precise plan" for scaling AI products. While the open-source Llama model is strategically valuable for ecosystem building, its direct monetization path remains unclear. Analysts expect Meta’s AI investments will eventually pay off through higher ad and subscription revenues, but the timing and strength of this chain are highly uncertain.
Amazon: AWS Growth Hits 15-Quarter High, Custom Chips Drive Differentiation
Among the three tech giants, Amazon arguably has the clearest logic for AI investment—at least based on disclosed data.
Amazon’s AWS Q2 revenue grew 28% year-over-year, with an annualized run rate of $150 billion, marking the fastest growth in 15 quarters. AWS operating margin hit a record 13.1%, validating Amazon’s core bet that "large-scale AI infrastructure spending will accelerate, not dilute, profitability." AWS quarterly revenue reached $37.6 billion, and backlog orders have swelled to $364 billion.
Amazon’s differentiator is its in-house chip strategy. Trainium2 is nearly sold out, and Trainium3 is almost fully booked. The custom chip business—including Trainium, Graviton, and Nitro—has annualized revenue exceeding $20 billion, with triple-digit year-over-year growth. OpenAI has committed to purchasing about 2 GW of Trainium capacity starting in 2027, and Anthropic has pledged up to 5 GW. Total Trainium revenue commitments now exceed $225 billion.
Financially, Amazon’s custom chip strategy directly impacts margins. CEO Andy Jassy says large-scale Trainium deployment will "provide hundreds of basis points of operating margin advantage." By replacing NVIDIA GPUs with lower-cost internal alternatives, AWS could push its operating margin back toward the 40% threshold if this advantage materializes.
Still, Amazon isn’t risk-free. The company plans to spend about $200 billion in capex in 2026. Q1 capex hit $44.2 billion, while operating cash flow was only $26 billion, resulting in negative free cash flow of $18.2 billion. Whether this cash burn rate is sustainable depends on AWS’s ability to convert growth into positive free cash flow in the medium term.
SK Hynix: Upstream Beneficiary and Valuation Challenge in AI
As an upstream supplier for AI infrastructure, SK Hynix’s performance is a key barometer for the health of the entire AI supply chain.
Consensus expects SK Hynix’s Q2 revenue to reach KRW 84.06 trillion, with operating profit at KRW 64.09 trillion—nearly 600% year-over-year growth, potentially setting a new record. Operating margin is projected to rise from 72% in Q1 to 75%–77%. If realized, SK Hynix will surpass TSMC in operating margin for the third consecutive quarter.
The main growth driver is HBM (High Bandwidth Memory), which is indispensable in AI data centers. As HBM capacity expands, supply for general-purpose memory remains constrained. The share of long-term contracts is rising, and sales to large tech companies and AI data centers are expected to reach 70%.
However, despite strong fundamentals, SK Hynix’s stock has dropped over 30% from its peak. This valuation discount reflects market concerns about a peak in the memory cycle and warnings from some brokers that Q2 results may "miss expectations." SK Hynix’s earnings are not just about the company itself—they’re a direct test of demand strength across the AI hardware supply chain.
Paradigm Shift in AI Investment Logic
Viewed through a unified lens, a clear trend emerges: AI competition is shifting from "who invests more" to "who can generate revenue."
In 2026, the four hyperscale cloud providers are expected to spend a combined $730 billion on AI infrastructure. Alphabet plans $195–$205 billion, Microsoft about $190 billion, Amazon about $200 billion, and Meta $125–$145 billion. Total capex is up roughly 77% from about $410 billion in 2025. Analysts expect this figure could break $1 trillion in 2027.
Yet investor confidence is waning. The 30-day correlation between major AI spenders and semiconductor stocks has plunged from +0.78 to nearly zero—the lowest reading in 4.5 years. Hyperscalers hold $1.65 trillion in off-balance-sheet AI obligations. All these numbers point to one conclusion: The market is no longer willing to pay a premium for the narrative of "potential future returns," but is demanding evidence of "returns happening now."
Boston Consulting Group’s 2026 AI Radar shows companies plan to double AI spending to about 1.7% of revenue in 2026, with 90% of CEOs pledging to keep investing even if returns take time. But the gap between commitment and earnings is the core uncertainty in current market pricing.
Conclusion
On July 29, Microsoft, Meta, Amazon, and SK Hynix will each deliver their results. The benchmark for these results is no longer the scale of capex, but its efficiency.
For Microsoft, the market needs to see Azure growth sustained around 40% and disciplined capex guidance. For Meta, the focus is whether strong ad revenue can offset AI-driven margin erosion. For Amazon, confidence in its $200 billion investment hinges on AWS’s growth and the execution of its custom chip strategy. For SK Hynix, record profits must reverse the stock’s downward trend—a direct test of AI hardware supply chain demand.
The long-term narrative for AI hasn’t changed—but market patience has. Billions in spending are now entering the validation phase, and the results will become clearer after July 29.
FAQ
Q1: Why is the market’s attitude toward AI spending changing?
The shift stems from the gap between surging investment and visible returns. In 2026, the four tech giants’ combined AI infrastructure spending is about $730 billion, up 77% from 2025. Meanwhile, the correlation between AI spenders and semiconductor stocks has dropped from +0.78 to nearly zero. Investors are no longer satisfied with the "scale of investment" narrative and are demanding verifiable evidence of "revenue output."
Q2: Where does Microsoft’s AI investment return show up most clearly?
Microsoft’s AI returns are evident in three areas: AI compute revenue within Azure cloud, enterprise Copilot subscription income, and the value-added impact of AI features on existing product lines like Microsoft 365. Microsoft’s AI business is running at a $37 billion annualized rate, with over 20 million paid Copilot seats. Since July 1, Microsoft 365 Business has integrated Copilot.
Q3: Why is Amazon’s custom chip strategy important?
Amazon’s Trainium custom chips can directly replace NVIDIA GPUs, delivering comparable inference performance at much lower cost. CEO Jassy says large-scale Trainium deployment offers "hundreds of basis points of operating margin advantage." Trainium’s total revenue commitments exceed $225 billion. Custom chips reduce dependence on a single supplier and are a key lever for AWS’s margin sustainability.
Q4: Why is SK Hynix’s performance so critical for the AI supply chain?
SK Hynix is a global leader in HBM, a key component for AI accelerators. Its results directly reflect real demand strength in AI data centers. The market expects SK Hynix’s Q2 operating profit to reach KRW 64.09 trillion, up nearly 600% year-over-year. If results meet expectations, it will validate sustained AI hardware demand; if not, it could trigger a repricing of the entire AI supply chain.
Q5: Among the three tech giants, whose AI investment logic is the clearest?
Based on disclosed data, Amazon’s AI investment logic is the most transparent. AWS growth is at a 15-quarter high, the custom chip strategy has generated quantifiable revenue commitments, and operating margin is at a record level. However, this also means Amazon faces the highest market expectations—if AWS growth slows or the chip strategy falls short, the stock could face the greatest adjustment pressure.
