Despite being the "global leader," Nvidia still faces a "valuation discount." Bank of America: It should strengthen "buybacks and dividends."

```

Nvidia’s valuation increasingly looks like a “cash distribution” issue, rather than just an AI cycle issue. Bank of America Securities shifts the perspective from computing power demand and product iteration back to a more fundamental question: When the peak of ecosystem investment passes, can Nvidia return its increasingly extraordinary free cash flow to shareholders in a more definite way, thereby triggering a valuation re-rating?

According to Hard AI, Bank of America Securities analyst Vivek Arya pointed out the core leverage of Nvidia’s valuation in his latest report—“Increasing returns helps expand the shareholder base and deliver a signal of sustained value.” He maintains a Buy rating for Nvidia, with a price target of $300—not because the market underestimates how much Nvidia can earn, but because investors are still hesitant about “how this money will ultimately be used, and whether it can be sustained.”

Nvidia is expected to generate more than $400 billion in free cash flow cumulatively in 2026-2027, a scale approaching the sum of Apple and Microsoft, yet the valuation is significantly discounted. The report argues thatif the company’s cash returns such as dividends/buybacks look more like those of “mature large tech stocks,” some concerns about growth durability and capital allocation can be hedged, and the valuation discount will have a better chance to converge.

However, dividends and buybacks aren't the only hurdles. The report also notes two restraining forces: first, Nvidia’s weight in the S&P 500 has reached about 8.3%, and the “additional buying space” for passive/benchmark funds is shrinking; second, competition is rising from AMD and other general GPUs, as well as Broadcom, Google, AWS, etc., on the ASIC path.

The crux of the valuation discount: not about earning too little, but “durability + capital allocation” not being priced in

The report starts from the point of “excessive discount.” By consensus estimate, Nvidia is currently trading at about 26x/19x P/E for CY26/27E, while the “Mag-7” average is about 49x/42x; the EV/FCF discount is even greater (Nvidia at about 28x/20x, with the group average even higher). On the PEG metric, Nvidia is around 0.36x, significantly lower than the group average of 2.61x.

The implication behind these comparisons is:the market does not doubt Nvidia’s short-term earnings explosiveness, but is more sensitive to two questions—can the growth be cross-cycle, and will management reinvest the cash flow in directions that shareholders find “uncertain” (such as unexpected large acquisitions, or more “noisy” supplier financing arrangements). Under this sentiment, even if the scale of cash flow is close to Apple + Microsoft, the valuation may continue to be suppressed.

Cash return as the “catalyst”: unlocks new shareholder structure

Nvidia’s current dividend yield is only about 0.02%, so its coverage in dividend/income funds is markedly low. Based on Lipper/EPFR data: among equity income funds, Nvidia is held by only about 16% of funds, versus a tech peer average of around 32%; the average peer dividend yield is about 0.89%.

In other words, Nvidia shares are naturally more concentrated among growth/momentum/benchmark funds. Once the market starts worrying about marginal changes in growth, this class of shareholders tends to be more “picky.” Higher, more predictable cash returns can open up some of the capital pools, making the shareholder composition more sticky and diversified, and closer to the path by which Apple and Microsoft accumulated a “long-term shareholder base” through their return policies.

“Enhanced” return plan: raise the dividend from symbolic to 0.5%—1%

Nvidia does not need to aggressively leverage or sacrifice investment; it is suggested to start with dividend improvement, the most visible approach:increase the dividend yield from 0.02% to 0.5%—1%, benchmarking Apple’s 0.4% and Microsoft’s 0.8% range.

It provides a “sufficient but not harmful” funding calculation: achieving the above dividend yield target would need about $26–51 billion, roughly 15%–30% of CY26E free cash flow, and 11%–21% of CY27E free cash flow. This means that even with a substantial increase in dividends, there is still room left for buybacks and ecosystem investment.

Over the past three years (CY22-25), Nvidia’s free cash flow return rate (dividends + buybacks/FCF) averaged about 47%, lower than peers at about 80%, and also lower than its earlier period (CY13-22) average of about 82%—the “low return” itself provides room for adjustment.

Two things dividends cannot resolve: excessive index position, more complex competition

For valuation suppression: Nvidia’s share of S&P 500 is about 8.3%, surpassing the historical peaks of Apple and Microsoft (7.9%, 7.2%). As the semiconductor sector’s share in the index also reaches about 17%, many “benchmark-hugging” funds find it objectively harder to continue adding positions. With the gradual listing of large private AI companies in the future, the index structure may see some “rebalancing,” with such constraints being relieved—but it won’t disappear immediately just because Nvidia pays more dividends.

Competition also applies: Nvidia currently faces two types of pressure—share grabbing from general chips like AMD, and substitution from Broadcom, Google, AWS, etc. via the ASIC/self-development route. It still expects Nvidia to maintain 70%+ AI value share, citing a more complete product portfolio, third-party-validated tokens/watt metric, over $95 billion in strategic prepayments locking up supply chain, ecosystem investment, and 100+ workload-optimized libraries and enterprise/developer adoption. But these advantages mainly determine “whether high profitability can be retained,” not directly “when valuation discount disappears.”

Taken together, the report’s stance is clear: the most difficult thing for Nvidia is not “proving the AI story again,” but to use more predictable cash returns to break open a gap in market discounting of growth durability and capital allocation. How wide that gap can be depends on whether it can simultaneously maintain competitive moats and distribute “earned money” more like the giants.

This article comes from WeChat public account “Hard AI”. For more cutting-edge AI news, please visit here

Risk alert and disclaimerThe market involves risk. Please invest cautiously. This article does not constitute personal investment advice and does not take into account individual users’ unique investment objectives, financial situation or needs. Users should consider whether any opinions, viewpoints, or conclusions herein are suitable for their specific circumstances. Investment is at your own risk. ```