Alphabet Says Its AI Spending Will Keep Rising. 2 Data Center Stocks to Buy and Hold Through 2030.

Alphabet spent $51.4 billion on capital projects in Q3 2026, more than double a year earlier, retained its $195 billion–$205 billion 2026 capex range, and expects spending to grow significantly in 2027. Nvidia reported fiscal Q2 revenue of $96.2 billion, up 106% year over year, and guided to about $108 billion for fiscal Q3, although sequential growth is slowing. Vertiv’s Q3 sales rose 61% to $4.3 billion, and it raised its 2026 adjusted EPS midpoint to about $7.15; the article favors Nvidia at roughly 14 times expected fiscal 2028 earnings versus Vertiv at about 26 times expected 2027 earnings, while noting Vertiv will stop reporting orders.

Analysis

AI infrastructure: spend is not the same as returns. The key transmission is whether installed capacity earns attractive utilization and pricing; if monetization lags, depreciation and power costs can pressure hyperscaler returns even while supplier revenue remains strong. That makes cloud growth, capacity utilization and customer returns on AI investment more useful confirmation than capex guidance alone.

For NVIDIA, the main second-order risk is substitution: Alphabet’s TPUs and other in-house accelerators can capture workloads that might otherwise use merchant GPUs. That need not displace NVIDIA outright, but it can limit pricing power and make spending growth a weaker proxy for NVIDIA revenue. Microsoft’s delivery timing also means quarter-to-quarter supplier demand may be lumpy rather than a clean read-through to end demand.

Vertiv has attractive operating leverage to data-center buildout, but reduced order disclosure raises the risk that investors discover a slowdown only through sales or margin revisions. Grid connections, equipment lead times and customer project delays could defer revenue without ending underlying demand.

Over the next 1–3 months, watch NVIDIA guidance and hyperscaler commentary on utilization and AI monetization; over 6–18 months, watch whether capex converts into cloud revenue and returns. The relative setup favors NVIDIA over Vertiv on the article’s cited earnings comparisons, but valuation gaps can persist and the businesses have different risk profiles. A broad capex pullback would hurt both; Vertiv’s weaker visibility makes it more vulnerable to a sentiment-driven de-rating.

Key Decisions for Investors

• Prefer a staged long in NVDA over VRT as the initial AI-infrastructure exposure; avoid chasing a single-day move and add only if subsequent guidance supports continued demand. The thesis is falsified by a material deterioration in NVIDIA’s forward revenue outlook or evidence that custom accelerators are taking share faster than expected.
• For accounts seeking a relative-value expression, consider long NVDA / underweight VRT rather than an unhedged VRT short. Size conservatively: the pair is not market-neutral, and a broad data-center spending acceleration could lift both. Reassess if Vertiv sustains sales and margin performance despite reduced order transparency.
• Keep GOOG and MSFT capex announcements on a watchlist, not as standalone buy signals. Seek confirmation in cloud growth, utilization and evidence of monetization; weaker returns on AI investment or delayed deployments would challenge the supplier thesis.
• Risk trigger: reduce exposure if hyperscalers guide to lower infrastructure spending, NVIDIA’s forward growth outlook weakens, or Vertiv’s sales/margins disappoint. Vertiv’s loss of order disclosure makes reported execution—not assumed backlog momentum—the key monitoring signal.