The Q1 2026 earnings season made one number impossible to ignore. Combined Mag 7 capex now runs above $500B for 2026 across sectors, with Amazon targeting $200B and Alphabet guiding to $180B-$190B. Investors holding only Mag 7 names are missing the second-order plays.
Why This Capex Cycle Matters for Investors Beyond the Mag 7
The picks-and-shovels framing is old, but the scale is new. Per Q1 2026 earnings coverage at The Next Web, Microsoft, Alphabet, Meta, and Amazon together committed over $650B in 2026 capex. Analyst estimates cited by The Motley Fool put AI infrastructure spending alone at $500B or more this year.
That money has to land somewhere: chips, electricity contracts, real estate, and cooling and networking gear. Each is a separate supply chain with listed beneficiaries. We mapped the broader landscape in our piece on AI stocks beyond NVIDIA; the four groupings below are the cleanest way to organize exposure.
Semiconductors: The Direct Beneficiaries
The first dollar of any AI capex budget goes to silicon. NVIDIA still captures most accelerator spend, but the supporting cast is where diversification lives. Broadcom sells custom AI ASICs to Google and Meta plus the networking chips inside training clusters. AMD is the credible second-source GPU. Arm Holdings sits one layer deeper, powering the host CPUs paired with GPU accelerators, including AWS Graviton.
The risk is concentration. If any Mag 7 name trims its 2027 capex guide, semiconductor multiples compress fast. The recent AI capex risk sell-off was a preview.
Power and Utilities: The Hidden Bottleneck
Electrons are now the binding constraint. The IEA projects global data center electricity use to roughly double by 2030, with AI workloads tripling their share. U.S. data centers already account for close to half of incremental electricity use, and utilities in PJM, ERCOT, and the Mid-Atlantic are signing decade-long supply contracts at premium prices.
The investable angle is broad. Regulated utilities in Northern Virginia, Phoenix, and central Ohio benefit from rate-base growth. Independent power producers with nuclear and gas baseload capture premium PPA pricing. Grid equipment makers and turbine OEMs sit on multi-year backlogs.
The risk is regulatory. Commissions decide who pays for grid upgrades, and pressure to shield residential ratepayers can compress utility returns.
Want to align your watchlist with the AI capex cycle? Audit your AI capex exposure across these four sectors at Gotrade App.
Data Center REITs: Owning the Real Estate
Hyperscalers do not build everything themselves. They lease meaningful capacity from operators that move faster on permitting and power. Equinix leads in interconnection-heavy colocation, where AI inference workloads sit close to enterprise customers. Digital Realty plays at the hyperscale end, with build-to-suit campuses leased to the same Mag 7 names funding this cycle.
The 2026 angle is pricing power. Vacancy in core markets has fallen below 3%, lease rates are stepping up double digits on renewals, and both REITs trade at premiums to NAV.
The risk is duration. REITs are rate-sensitive, and a sharp move in long-end yields would pressure the group even with fundamentals intact.
Cooling and Networking: The Specialized Picks
The last leg is the least glamorous and arguably highest beta. AI clusters generate heat at densities air cooling cannot handle, so liquid and direct-to-chip systems are now standard. Vertiv is the pure-play, supplying power management and liquid cooling that show up in nearly every hyperscale RFP.
Arista Networks dominates the high-speed Ethernet switches connecting GPU clusters at Microsoft, Meta, and other AI-heavy buyers. As clusters scale from thousands of accelerators to hundreds of thousands, the switching fabric becomes a bigger share of the bill of materials.
The risk is competitive. Both face well-funded entrants, and any sign hyperscalers are pulling cooling or switching in-house would dent the multiple quickly.
Conclusion
A $500B-plus AI capex year lifts every link in the supply chain that enables the spend. Semiconductors, power, REITs, and cooling and networking are the four cleanest baskets to capture that flow.
The right move for most investors is spreading exposure across the cycle, not picking one winner. Holding only Mag 7 names concentrates you in the spenders. Adding picks-and-shovels names broadens you across suppliers, landlords, and utilities that get paid regardless of which hyperscaler wins the next model race.
Review your AI capex exposure at Gotrade App.
FAQ
How much are the Mag 7 spending on AI in 2026?
Combined hyperscaler capex is tracking above $650B for 2026, with roughly $500B of that earmarked for AI infrastructure.
Which sector benefits most directly from AI capex?
Semiconductors capture the largest share of each capex dollar, led by NVDA, AVGO, AMD, and ARM IP licensing.
Why are utilities part of the AI trade?
Data centers are projected to roughly double global electricity use by 2030, forcing utilities to sign long-dated supply contracts at premium prices.
What is the biggest risk to picks-and-shovel stocks?
A meaningful cut to forward Mag 7 capex guidance would compress multiples across all four sectors quickly.
