Big Tech Will Spend $745 Billion on AI This Year. These Stocks Get Paid on Every Dollar.
Amazon, Alphabet, Microsoft, and Meta Platforms are on track to spend roughly $745 billion on capital expenditures this year, according to The Motley Fool’s research. Most of this spending is going toward data center infrastructure, but it also includes other non-artificial intelligence (AI)-related expenditures.
Nvidia (NVDA -3.16%), Broadcom (AVGO -3.10%), and Arista Networks (ANET -4.45%) supply the compute, custom silicon, and networking that turn that capital into revenue from working AI cloud capacity. These suppliers sit at the center of the AI boom, with every dollar spent on data centers ultimately padding their coffers.
Image source: Getty Images.
Nvidia remains the default choice for data centers
Nvidia designs graphics processing units (GPUs) and related computing systems that power AI training and inference workloads in the data center. The growing use of Broadcom’s custom-designed chips has threatened Nvidia’s dominance in the AI chip market. Still, Nvidia maintains a solid competitive position due to its CUDA software platform, which has more than 7.5 million users worldwide, and an extensive supply chain built up over decades.
Today’s Change
(-3.16%) $-7.22
Current Price
$221.30
Key Data Points
Market Cap
Day’s Range
$219.64 – $226.00
52wk Range
$164.27 – $243.37
Volume
1.8M
Avg Vol
118.9M
Gross Margin
74.67%
Dividend Yield
0.23%
Nvidia benefits from unmatched scale and relationships with component suppliers, giving it priority access when demand surges. Its combined offering of software and hardware, such as the new Vera Rubin platform, provides an advantage in selling to enterprise customers that custom chipmakers can’t easily replicate.
In the second quarter of fiscal 2027 (ending in January), total revenue more than doubled year over year to $96 billion, mostly driven by surging demand in the data center market. Management expects 70% revenue growth in fiscal 2028.
Custom chips may continue to take a growing share away from Nvidia’s GPUs in the data center market. Still, the GPU leader is mitigating this threat by expanding its relationships with non-hyperscaler customers, including enterprises and other industrial opportunities. Revenue from these markets grew 138% year over year last quarter, reaching $40 billion. The stock trades at a forward price-to-earnings (P/E) multiple of roughly 14 times based on next year’s consensus earnings estimate, which looks attractive against this momentum.
Today’s Change
(-3.10%) $-11.06
Current Price
$346.25
Key Data Points
Market Cap
Day’s Range
$342.82 – $354.25
52wk Range
$289.96 – $495.00
Volume
221K
Avg Vol
22.3M
Gross Margin
66.54%
Dividend Yield
0.75%
Broadcom’s custom silicon is ramping up with the largest AI customers
Broadcom’s AI chip revenue is growing much faster than Nvidia’s, up 221% year over year in the fiscal third quarter of 2026. It designs custom AI accelerators such as Google’s Tensor Processing Units (TPUs), OpenAI’s Jalapeño, and Meta’s MTIA, as well as high-speed networking components.
Broadcom’s growth trajectory is closely tied to the ongoing improvement of the most advanced AI models from Anthropic, OpenAI, Meta, and Google, which require high-performance custom silicon. Broadcom’s CEO Hock Tan said AI revenue is expected to grow from $58 billion this year to $230 billion by fiscal 2028.
The main risks for Broadcom stem from the momentum of the current AI boom and anything that slows it, such as regulations on data center construction. Still, the stock is already trading at a modest forward P/E of 18 times next year’s earnings estimates, which seems to account for potential hurdles. Given the AI revenue outlook for the next few years, the stock still offers attractive upside prospects.
Today’s Change
(-4.45%) $-9.10
Current Price
$195.42
Key Data Points
Market Cap
Day’s Range
$193.30 – $203.68
52wk Range
$114.52 – $217.41
Volume
13.6K
Avg Vol
6M
Gross Margin
63.00%
Arista is a top Ethernet fabric supplier for data centers
Arista Networks supplies interconnects and other networking gear needed to connect all the chips deployed in data centers. It sells high-performance Ethernet switches. Its EOS operating system, which runs the switches and routers, makes it more costly for customers to switch to another networking supplier and is therefore a key competitive advantage.
The company’s quarterly revenue just surpassed $3 billion for the first time, growing 38% year over year in the second quarter of 2026. Management raised full-year revenue guidance to $12.6 billion, representing 40% growth over 2025. Arista’s high-speed networking systems, or AI fabrics, now span more than 100 companies, which would include the big hyperscalers, specialized cloud providers (neoclouds), AI labs, and enterprises.
Risks include deployment delays as AI data centers become more complex. This could require lengthy testing and acceptance before a customer can deploy its networking systems in their data centers. Arista said its deferred revenue was about $6.9 billion in Q2 2026.
Still, management said on its August 2026 earnings call that it continues to win new customers, ramp up new products, and expand into new use cases for its products. A lot of growth is already baked in, with the stock trading at roughly 38 times next year’s earnings estimates — a steeper premium than Nvidia and Broadcom. The value may lean more toward leading chip stocks than a pure-play networking player right now.
