2 October 2025
Artificial intelligence has ignited one of the largest capital spending waves in history, with listed companies committing more than $1 trillion to AI-related infrastructure, chips, and platforms. But while this flood of investment makes headlines, the real story lies in who actually benefits from the spending. From semiconductor giants powering the compute revolution to cloud operators building data centers at breakneck speed, and even utilities racing to meet soaring energy demand, the ripple effects of this AI gold rush extend far beyond the tech titans writing the checks.
Beneficiary categories & mechanics
Here are how the gains and cash flows tend to be distributed, and who tends to capture those gains:
| Beneficiary type | Why they benefit | Risks / constraints | Examples or sub-segments |
|---|---|---|---|
| Semiconductor / chip designers & foundries | AI workloads (training, inference) demand massive compute, especially GPUs, TPUs, ASICs, memory (DRAM, HBM, SRAM) etc. So increased demand raises volumes, pricing power, and innovation premiums. | Supply constraints, capital intensity, process yields, competition, geopolitical / export controls | Nvidia (GPUs / AI accelerators), AMD, TSMC, Samsung, Micron, SK Hynix |
| Semiconductor equipment / tooling / EDA | To make next-generation chips, you need lithography, etching, deposition, AR/VR, etc. Tooling providers get revenue from more advanced chip manufacturing. | Long cycles, high R&D investment, client concentration | ASML, Applied Materials, Lam Research, Cadence, Synopsys |
| Cloud / data center operators / infrastructure / colocation | Many companies will outsource AI workloads (training, inference) to cloud providers; also, data centers and colocation facilities must scale. | Power, cooling, network bandwidth, cost of capital, differentiation | Microsoft Azure, Amazon AWS, Google Cloud, Oracle Cloud, Equinix, Digital Realty |
| Networking / interconnect / storage / switching / memory / cooling / power | AI adds stress on data flows, requiring faster networks, more storage, faster interconnect, cooling systems, power infrastructure. | Bottlenecks in supply chain, energy constraints, regulatory challenges | Broadcom, Arista Networks, Cisco, Super Micro, storage firms like Western Digital, memory firms, cooling / HVAC firms |
| Software / AI platforms / middleware | Once the infrastructure exists, AI model providers, software stacks, inference frameworks, APIs, services capture value via licensing, subscriptions, custom solutions. | Model differentiation, data access, compute cost, competition | OpenAI / model providers, Microsoft’s AI tools, software firms embedding AI, smaller AI service providers |
| Utilities / energy / power / grid infrastructure | Data centers consume huge amounts of electricity. To enable expansion, the grid, renewables, power providers, battery / storage firms benefit. | Regulation, grid capacity, energy cost and carbon constraints | Utility companies, renewable energy firms, battery / storage firms |
| Real estate / REITs for data center / land / infrastructure | Properties and underlying land for data centers, especially in favorable zones, benefit from the real estate demand. | Zoning, latency, location constraints, capital intensity | Data center REITs, land development near fiber / power corridors |
| Suppliers / peripheral / component vendors | Cable, fiber, connectors, power supplies, cooling systems, specialized sensors, memory modules, etc. | Component commoditization, margin pressure | Smaller niche hardware suppliers, contract manufacturers, power electronics firms |
| End users / adopters (productivity gains) | Many firms investing in AI will capture cost savings, margin expansion or new revenue streams via automation, better analytics, etc. — so productivity gains flow to those deploying AI well. | Execution risk, integration challenges, regulatory / ethical risks | For instance, large industrial, logistics, pharma, financial services firms that adopt AI successfully |
| Shareholders in the investing / sponsoring companies | If a company invests in AI infrastructure (or equity stakes in AI firms), it may internalize and capture returns if it has competitive differentiation. | Capital risk, execution risk, obsolescence risk | Large tech firms like Microsoft, Nvidia, Google, Meta often both investors and infrastructure owners |
| Labor / human capital (in cases of new jobs, reskilling) | Some roles benefit from AI augmentation, new AI-adjacent roles, training, etc. Though this is more contested given automation risks. | Displacement, skill/education mismatch, social backlash | Educators, AI trainers, specialized engineers, data scientists |
Relative magnitude & empirical evidence
Here’s what research and market observations suggest:
- Semiconductors and infrastructure are early and “safe” picks: Analysts and research firms (like Goldman Sachs, VettaFi) emphasize that the “next phase” of AI investment will disproportionately favor infrastructure players — semiconductors, cloud providers, network & equipment makers, REITs, utilities, security software etc.
- Survey / CIO signals point to Microsoft, Nvidia: CIO surveys reveal that enterprises are prioritizing AI infrastructure purchases (e.g. GPUs, cloud services). Microsoft and Nvidia often come out as beneficiaries of increased spend.
- Supply chain constraints / bottlenecks: For example, demand for memory and storage is surging. Semiconductor equipment firms are stretched. The complexity and scarcity of advanced process nodes, EUV tooling, etc., give providers of those tools leverage.
- Productivity / margin capture estimates: Morgan Stanley estimates that, for large public companies that adopt AI, there’s potential for significant net gain (after costs) — i.e. companies themselves as adopters also capture value.
- Surplus-allocation model: Morgan Stanley’s “AI Beneficiaries” work (in their public and internal research) tries to allocate how much of the gains from AI accrue to suppliers, how much to adopters, how much to customers. They estimate that of the cost savings / efficiency, a portion is captured by suppliers, a portion by customers, a portion by the adopting companies themselves.
Thus, not all of the AI investment “pays off” equally — some of it is absorbed by capital expenditures, operating expenses, depreciation, and competitive price pressures.
Who is least likely to benefit
To better sharpen your lens, it helps to see who tends not to benefit (or benefit less):
- Commoditized hardware: Low-margin commodity parts often see margin pressure.
- Incumbent firms in legacy industries that fail to adopt or integrate AI: They may be disrupted rather than benefiting.
- Regulated sectors with limited capital flexibility (e.g. some utilities, telecoms in rigid regulatory regimes) may struggle to capture the upside or pass costs through.
- Small vendors without scale, R&D, or access to capital might be squeezed.
Synthesis: Who are the main beneficiaries today?
If I had to pick the top “bet” buckets (by magnitude and durability) in the AI investment wave, these would be:
- Leading semiconductor / AI accelerator firms (Nvidia is a poster child)
- Cloud / hyperscalers / data center operators (they host and monetize AI workloads)
- Chipmaking & equipment / EDA tool providers
- Network / interconnect / storage & cooling / power infrastructure firms
- Software / AI platforms & service providers
- Utilities / energy providers (especially those upgrading grid capacity or co-locating renewables at data centers)
These are the slices of the value chain where price leverage, technological differentiation, and scale advantages are most durable.
For investors, the lesson is clear: not every company pouring billions into AI will see its stock outperform. The real beneficiaries are those sitting at the choke points of the value chain—chipmakers, cloud infrastructure leaders, data center REITs, power suppliers, and specialized software enablers. Positioning your portfolio around these segments allows you to capture the durable earnings growth fueled by AI’s trillion-dollar spending wave, while avoiding the risk of backing firms that may overspend without a clear path to monetization. In a market defined by hype and capital intensity, disciplined exposure to the beneficiaries—not just the spenders—will separate strong long-term portfolios from the rest. Let’s ride this Bull Run and finish the last quarter of 2025 strong!
Ben Evangelista & Partners 