Some analysts say that semiconductor giant Nvidia (NVDA +1.29%), whose chips have formed the backbone of the AI revolution so far, should be worried about Alphabet's (GOOGL +0.42%) $900 billion project around tensor processing unit chips.
Others say that the OpenAI partnership with Advanced Micro Devices (AMD +3.17%) is suddenly threatening Nvidia's 90% market share for graphics processing units. Still others view Nvidia's position as unassailable, with Bank of America analyst Vivek Arya stating the company operates in a "different galaxy" than its competitors.
Yet many of us are old enough to remember when Nvidia stock lost a sixth of its value in January 2025, as China's AI chatbot DeepSeek rattled markets, with Wall Street terrified that an AI app made at a fraction of the cost of U.S. rivals could upend the industry.
Image source: Getty Images.
This will be another year of record-breaking AI infrastructure spending, with the "hyperscalers" (companies building AI data centers at massive scale) forecast to spend $452.7 billion, according to Morningstar. Their combined capital expenditures will be four times greater than what the publicly traded U.S. energy sector spends to drill oil, refine and transport gasoline, and run large chemical plants.
And here's one stunning fact: In the age of AI, Amazon's capital expenditures are greater than that of America's entire energy sector combined.
With multi-trillion-dollar companies sparing no expense to win the AI boom, it's impossible to know which will rise or stumble in the looming wave of disruption. And apart from being unknowable, the question of who wins the battle for market share may be the wrong question to ask. At the end of the day, the AI revolution and the $15.7 trillion in wealth it could create may be big enough for everyone.
A catch-all fund to play the $15.7 trillion wealth wave
For investors seeking exposure to some of the most entrenched leaders of the AI revolution, the VanEck Semiconductor ETF (SMH +3.29%) is a straightforward play.
The exchange-traded fund, with its $37.6 billion in total net assets, is designed to track the returns, as closely as possible before fees and expenses, of the 25 largest and most liquid semiconductor firms. And since ChatGPT ushered in the age of AI in late 2022, its returns have been impressive.
Over the last three years, the fund has achieved an average annual return of 46.83%. This return leaves the tech-heavy Nasdaq Composite, with its three-year average annual gain of 31.12%, in the dust.
NASDAQ: SMH
Key Data Points
Alas, these gains are in the past, and aren't reliable indicators of how the fund could perform going forward. But there are three reasons I like the VanEck Semiconductor ETF as a catch-all play to tap into the wave of AI-driven wealth creation without betting everything on one or two volatile stocks.
First, the fund is diversified, with its largest holding, Nvidia, making up 21% of its portfolio. Its next largest holding, Taiwan Semiconductor Manufacturing, makes up just over 10% of the portfolio, while none of its other top 10 holdings crack double digits.
Because its methodology allows portfolio selection for both domestic and U.S.-listed foreign companies, the fund is geographically diversified as well. However, it is oriented toward large-cap companies, as it seeks to mirror the returns of the largest and most established semiconductor firms. This doesn't bother me, as I'm more interested in deep-pocketed firms that can spare no expense in protecting their industry leadership and even gain footholds in new markets.
Second, the fund offers at least a partial solution to investors who fear overvaluation in AI companies. The average price-to-earnings (P/E) ratio of its holdings, weighted as a percentage of its portfolio, amounts to 41.3. That means its shares trade at a premium relative to the broader market, but when you consider how fast its top holdings are growing earnings (at 65% year-over-year in Nvidia's case) it's far from the "nosebleed valuation" P/E ratios of around 200 seen at the eve of the dot.com bubble collapse.
Thirdly, the fund carries a modest expense ratio of just 0.35%, compared to the industry average of 0.56%. The below-average fee means that for every $10,000 invested, just $35 goes toward management fees.
Given the fund's stellar performance, both over the last three years and since its 2011 inception, in which shares have returned 26.8% on average each year, this management fee seems reasonable, to say the least. For investors who prioritize capital appreciation, diversification, and a simple and effective way to play the ongoing AI boom, the VanEck Semiconductor ETF is worthy of consideration.
