AI build-out fuels chip and data center bets

Artificial intelligence remains a major investment theme in 2026, with spending on AI infrastructure set to reach new highs. The five largest hyperscalers are expected to invest roughly $700 billion in capital expenditures on AI data centers this year, an amount larger than the gross domestic product of all but about 24 countries.

Yet market sentiment toward AI stocks has turned more selective. Investors see hyperscalers committing record sums to capital improvements while the financial returns from those projects are still uncertain. Some large cloud providers have begun issuing debt to support these plans, and their AI-related capex is now at levels that push up or exceed operating cash flow.

Despite this caution, research cited by The Motley Fool indicates that AI usage among businesses remains relatively low. Only about 18% of businesses are currently using AI, a figure expected to rise to 22% in the coming months, while larger firms show a 27% usage rate. This gap between infrastructure investment and real-world deployment underpins many of the current debates around an AI-driven market bubble.

AI computing resources remain constrained even with today’s relatively modest enterprise adoption. McKinsey & Company projects that about $7 trillion in data center capital expenditures will be required by 2030 to meet AI computing demand. For comparison, AI hyperscalers are expected to spend about $650 billion this year, leaving a sizable shortfall versus that long-term projection.

Goldman Sachs (GS) estimates that data center demand will exceed available supply by an average of 10 gigawatts over the next three years. This imbalance is encouraging specialized providers to expand capacity quickly and is shaping capital allocation decisions across the AI ecosystem.

Nvidia (NVDA) sits at the center of the current build-out as the primary supplier of graphics processing units used in AI data centers. The company now accounts for more than 7% of the S&P 500 (SPX) index, and its valuation is described as reasonable at a forward price-to-earnings ratio of about 22, contrasting with the far higher multiples seen in earlier tech booms.

In one notable move, Nvidia (NVDA) plans to invest $2 billion in Nebius Group, a neocloud infrastructure provider focused on dedicated AI data centers. Nebius builds AI facilities equipped with Nvidia systems, renting capacity to customers and offering managed software services on a full-stack platform.

Nebius added 170 megawatts of data center capacity in 2025, beating its 100 MW target, and expects 800 MW to 1 gigawatt of active capacity by the end of 2026. With Nvidia’s backing, the company aims to secure 3 GW of contracted power by 2026 and to bring more than 5 GW of data centers into operation by 2030. Analysts currently forecast Nebius’ 2026 revenue to rise 531% to $3.35 billion.

Beyond Nvidia, several hardware suppliers are benefiting from AI-related spending. Taiwan Semiconductor Manufacturing is highlighted as a key beneficiary because it produces most of the advanced chips used in AI applications. Taiwan accounts for 60% of the world’s semiconductor output and 90% of advanced chips of 7 nanometers or smaller, with TSMC holding 72% foundry market share as of the third quarter of 2025.

TSMC manufactures Nvidia’s GPUs, as well as chips for Advanced Micro Devices (AMD), Apple (AAPL), Broadcom (AVGO) and Qualcomm (QCOM). The company reported 2025 revenue of $122.4 billion, up 35.9% year over year, with diluted earnings per share rising 46.4% and a net profit margin of 45%. It projects 30% revenue growth in 2026 and a 25% revenue compound annual growth rate through 2029.

Memory supplier Micron Technology (MU) is also emerging as a major AI beneficiary. The company has become a leader in high-bandwidth memory, which is critical for advanced AI workloads. In the first quarter of fiscal 2026, Micron’s (MU) revenue rose 56% year over year to $13.6 billion, following a 49% increase in fiscal 2025, while net income grew from $1.9 billion to more than $5.2 billion.

The scale of AI investment has raised questions about a potential stock market bubble, particularly if data center spending peaks or slows. Commentators note that many hyperscalers, including Alphabet (GOOGL), Amazon (AMZN), Microsoft (MSFT) and Meta Platforms (META), could eventually benefit if front-loaded capex allows them to cut spending later and expand free cash flow.

Some analysts argue that even if an AI infrastructure bubble were to form, it might shift market leadership rather than undermining the entire market. They emphasize that AI demand is still in its early stages given low enterprise adoption, while capital continues to flow into core chip, foundry and data center providers.