Since Nvidia is already the world’s largest company in terms of market capitalization, many investors assume that their buying opportunity has passed.

But Will McGough, chief investment officer at Prime Capital Financial, doesn’t think it’s too late. His case rests on Nvidia’s price-to-earnings ratio, which compares a stock’s price with its earnings per share, and on the company’s dominant position in the field of artificial intelligence. He argues that a large market cap does not necessarily indicate that a stock is expensive.

That distinction matters for investors trying to decide whether to add to their artificial intelligence holdings, diversify them, or simply hold onto what they already own. According to McGough, there are a few indicators to keep an eye on, such as a leveling off in artificial intelligence spending, or the rise in 10-year yields toward the mid-5% range or near 6%, both of which could pressure equities broadly.

Here’s why he believes Nvidia is a great long-term buy — even at current prices.

Why Nvidia’s size is not the same as its valuation

A company’s market capitalization is the total value the stock market assigns to its outstanding shares. On its own, however, it tells us very little about how much investors are paying for each dollar of profit. The main point McGough makes is that Nvidia’s valuation should be assessed against its earnings rather than dismissed because the company is already so large.

It’s the biggest company by market capitalization in the world, and it’s undervalued relative to the market in terms of PE ratio.

McGough believes that, if NVDA’s earnings per share grow while its price-to-earnings ratio stays unchanged, its stock price would generally find support from those higher profits. But a falling valuation multiple could also offset earnings growth — that’s why he stresses that investors should consider both sides of the equation.

Nvidia is so attractive precisely because of its role in the infrastructure needed for artificial intelligence. GPUs, or graphics processing units, are chips used for the intensive computing behind many artificial intelligence workloads. CPUs, or central processing units, handle broader computing tasks. According to McGough, Nvidia participates in both arenas as companies continue to spend on data centers and related hardware.

Investors considering Nvidia should ask themselves whether their assumptions about future earnings, capital spending, and valuation are consistent with the price they are paying — regardless of the price.

How investing in Nvidia is also an investment in its startups

McGough also sees Nvidia’s investments in startups and private companies as part of its appeal, since investors usually cannot directly buy stakes in private companies. Owning Nvidia, he reasons, could provide indirect exposure to the businesses that it invests in or acquires as it builds out its artificial intelligence ecosystem.

Access is the new alpha.

McGough’s belief runs counter to viewpoints that investors need a dedicated private-market allocation to gain exposure to early-stage companies. His reason is that a public company with considerable resources is able to make investments that public shareholders otherwise could not.

Investors should also separate this investment theme from Nvidia’s chip business. According to McGough, the chip business, data center demand, and corporate investments may reinforce one another, but they are not the same source of returns. With that in mind, he affirms that investment decisions for Nvidia should be based primarily on the company’s overall earnings as well as its valuation.

Why the AI buildout includes TSMC, AMD, and NVDA

McGough’s stock picking discussion isn’t relegated to NVDA alone. He pairs Nvidia with TSMC, the manufacturer of chips used across the industry, and Advanced Micro Devices, which he describes as having greater exposure to cloud infrastructure and CPUs, in order to capture different parts of the capital spending cycle.

TSMC serves as an example of McGough’s wider criticism of placing too much reliance on traditional investment classifications. He maintains that investors should look beyond such categories as growth investing, value investing, and investments in emerging or developed markets to get a proper understanding of the businesses involved.

Although TSMC is based in Taiwan and therefore may be included in investments in emerging markets, McGough mainly sees it as a manufacturer situated at the heart of the demand for artificial intelligence hardware.

He believes artificial intelligence has become a meaningful part of emerging markets exposure through TSMC, SK Hynix, and Samsung. That may be counterintuitive for investors who associate emerging markets mainly with commodities or country-level economic growth.

The implication, in McGough’s view, is that an investor seeking chip exposure may need to look outside the United States instead of relying only on domestic technology stocks.

AMD represents a separate decision for investors who worry they have missed its run-up. McGough said investors trying to make short-term tactical trades may wait to see whether the stock cools off. For long-term investors, he emphasizes establishing appropriate exposure and accepting that a stock tied to a fast-moving theme often comes with volatility.

Above all, he wants investors to consider their investment horizon. A shorter-term trader will focus on the price paid and the possibility of a pullback. But a long-term, buy-and-hold investor decides whether the position fits a diversified asset allocation, or the mix of stocks, bonds, and other holdings designed around financial goals and risk tolerance.

Why capital spending is the key risk for AI hardware

While McGough is bullish on AI, he doesn’t assume that spending on artificial intelligence infrastructure will increase indefinitely. He says investors should watch capital expenditures, commonly called capex, which refers to corporate spending on long-lived assets such as data centers, servers, and equipment. That means that if the largest technology companies slow their spending, manufacturers and hardware suppliers could feel the effects.

This is especially relevant to TSMC and the broader semiconductor supply chain. Spending by hyperscalers, the large cloud-computing companies that operate vast data centers, supports orders for chips and related infrastructure. So, a slowdown in that spending could flatten demand — even if artificial intelligence remains an important long-term technology trend.

McGough, therefore, favors exposure to both hyperscalers and hardware-related companies rather than one or the other.

Geopolitics and currencies are additional concerns for McGough. He hasn’t formulated a numerical threshold to watch for either risk factor, so that means investors should avoid treating the broad artificial intelligence narrative as a substitute for evaluating company-specific and country-specific risks.

How higher interest rates could challenge AI stocks

The biggest shared risk McGough named for his six stock picks was an “out-of-control rate market.” His focus was on the 10-year yield, which is the return investors demand to own a 10-year U.S. government bond. He says a move into the mid-5% range and approaching 6% could cause equity markets to “break,” while acknowledging that investors do not know exactly when market conditions will change.

Higher yields matter to technology stocks because investors use interest rates when evaluating the present value of future expected profits. McGough did not lay out a valuation model, but his warning is clear. A higher-rate environment can also offer more appealing alternatives in fixed income, or bonds and similar debt investments.

For investors in or near retirement, McGough said investment-grade bonds and high-yield debt may help meet financial-plan return targets when yields are attractive. Investment grade refers to debt issued by borrowers considered more likely to repay, while high yield refers to lower-rated debt that generally offers higher income in exchange for greater credit risk. The appropriate mix depends on each investor’s circumstances, he says.

That creates a genuine trade-off. Investors who want the long-term growth potential of artificial intelligence stocks must be prepared for volatility and the possibility that rates pressure valuations. But investors who need income or have less capacity for stock-market swings may find that fixed income plays a larger role in their portfolios when yields are higher.

How to stay diversified when artificial intelligence is basically everywhere

McGough says that clients often ask him how to diversify away from artificial intelligence. His answer is that artificial intelligence increasingly touches many parts of the economy, including financial services. Still, he highlights JPMorgan Chase JPM and financial companies as one way to add exposure to businesses with a different economic driver than chip sales and data-center construction.

His reasoning is that higher rates can help major banks earn wider spreads on lending, while firms such as JPMorgan Chase also have asset management, private banking, and investment-banking operations.

Eli Lilly LLY offers another distinct source of exposure in McGough’s list. He cites GLP-1 drugs, describing Eli Lilly as the only non-technology company among the largest S&P 500 companies by market cap. He also identifies regulation, pharmacy pricing, and public concern about health care costs as risks that investors should weigh before jumping in.

McGough believes diversification should be judged by the economic forces behind each holding, rather than by a simple count of stock tickers.

Nvidia, TSMC, AMD, JPMorgan Chase, and Eli Lilly all have different businesses, but each could be affected by higher interest rates or changes in economic growth. So investors should consider how each new position affects their existing portfolio instead of assuming that just because a company belongs outside technology, it will automatically provide diversification.

The takeaway for Nvidia investors

McGough asks investors to look past NVDA’s market cap and focus instead on earnings, valuation, and its role in the artificial intelligence buildout. He believes rising earnings could support stock prices if price-to-earnings ratios do not fall enough to offset the gains. His conclusion is that Nvidia is undervalued, and his opinion is one that investors should test against their own assumptions.

To do so, they can ask themselves three questions: First, do they already have substantial exposure to artificial intelligence through individual stocks, funds, or broad market indexes? Second, can they tolerate a decline if capital spending slows or interest rates rise? Third, would Nvidia’s potential role in long-term growth improve their portfolio’s overall asset allocation, or merely add to an existing concentration?

Depending on how investors answer those questions, they may decide that Nvidia fits into a long-term, buy-and-hold portfolio. But those who need income, have a shorter time horizon, or already have concentrated artificial intelligence exposure may reach a different conclusion. McGough’s larger point is that the answer lies on earnings, valuation, risks, and portfolio fit — and not on the assumption that a company is either automatically too expensive or automatically “safe” simply because it is a market leader.

McGough selects Nvidia as his preferred stock for the next five years and also says it would receive the largest share of a hypothetical $10,000 allocation among his six picks.