KPMG He noted that graphics processing units, or GPUs, long associated with gaming and high-performance computing, are quickly becoming the cornerstone of AI infrastructure. Their parallel processing capabilities make them indispensable for training complex models, executing inference tasks, and supporting data-intensive applications across industries. As demand for AI grows, GPUs are shifting from dedicated hardware to an alternative investment category.
Sustained structural demand is supported by chronic supply shortages and income-generating rental arrangements. high net worth investors It’s a way to directly participate in the AI economy.
KPMG’s latest analysis, the third in a series examining GPU dynamics, draws on an exclusive survey of 120 participants. investment professionals and high net worth individuals.
Reveals a strong appetite technologythemed alternatives; 31 percent of respondents allocated capital to technology assets in the last three years; ahead of real estate (21 percent) and private equity or venture capital (20 percent).
The dominant motivations of those interested in GPUs are clear: 70 percent want meaningful capital appreciation, while 54 percent value portfolio diversity beyond traditional markets.
Innovation attractiveness (47 percent) and long-term presence Protection (46 percent) also has a distinctive feature.
Current allocations remain modest, reflecting the relative newness of the asset class.
Although the risk rate of many portfolios is below 5 percent, forward-looking expectations signal growth.
Nearly half of respondents anticipate significant scaling up of commitments in coming years, and optimism is high: 51 percent describe their outlook as very positive, and 24 percent describe it as somewhat positive.
Pricing trends reinforce this confidence. NVIDIABlackwell series GPUs have increased by 15-23 percent in recent months; Island models, on the other hand, have increased by 5-10 percent, and delivery times have extended to three to seven months due to fierce competition from hyperscalers, enterprises and governments.
Leasing models provide the return component that appeals to infrastructure-style investors.
private equity funds and specialized operators are increasingly funding GPU clusters, colocation facilities, and “GPU as a service” platforms. Similar themes appear in reports from other leading companies.
McKinsey highlights a multitrillion-dollar race to expand artificial intelligenceReady-made data centers where GPU operators, colocation providers, and hyperscalers derive value through utilization rates and power-efficient designs.
Deloitte He notes that AI chip sales could reach hundreds of billions by 2027, citing the fastest GPU adoption for applications such as financial services fraud detection.
Goldman Sachs It highlights the rise of “neocloud” providers building dedicated GPU capacity, as well as hyperscaler commitments approaching $1 trillion by 2027.
Risks continue to be at the forefront. sixty one percent KPMGTheir respondents consider GPUs to be slightly higher risk than typical alternatives; variability in hardware prices, rapid technological obsolescence (chips typically have an economic life of four to six years), regulator review, energy consumption and liquidity constraints in emerging vehicles.
Barriers to wider adoption include limited awareness (58 percent) and the need for clearer benchmarks and ESG-compliant structures.
Market maturation continues. Investors are demanding more standardized products.ETFspartnerships with private funds and established financial institutions to increase accessibility and trust.
Ace Artificial Intelligence penetrates As the complexity of each sector and model continues to increase, GPUs are expected to evolve into a stable, diversified holding, similar to digital infrastructure or renewable energy assets. GPUs represent more than hardware; They form the physical backbone artificial intelligence evolution.
