``` The increase is far below the chip index! Bank of America says Nvidia's "valuation is unreasonable" ```

``` The increase is far below the chip index! Bank of America says Nvidia's "valuation is unreasonable" ```

The market is mispricing, and Nvidia’s unjustifiable “discount” is a golden buying opportunity. Although Nvidia dominates the AI field, its stock price has only risen 3% year-to-date, far lagging behind the 82% surge in the Philadelphia Semiconductor Index. Bank of America believes this huge divergence reveals an extremely attractive buying opportunity.

According to Wind Chasing Trading Desk, the core logic of Bank of America’s (BofA Securities) July 7th research report is clear: The current valuation discount of Nvidia stems from the market’s overpricing of memory costs, ASIC competition, and position concentration, rather than any real deterioration in fundamentals.

The bank believes the market is currently pricing in an “unreasonable discount” of as much as 30-35% to Nvidia’s expected EPS in 2027/2028, bringing its forward P/E ratio down to a seven-year low (18x). In other words, the market is paying for a risk that doesn’t actually exist.

At the same time, BofA notes that the market has overestimated concerns about rising memory costs, custom chip (ASIC) competition, crowded institutional positions, and capital efficiency, while completely ignoring Nvidia’s unmatched pricing power, $119 billion in supply chain commitments, and continuously expanding market share.

With the AI data center capex supercycle not yet peaking and Nvidia’s moat deepening, BofA believes the current stock price offers an “enhanced buy opportunity,” reiterates a “Buy” rating on Nvidia, and sets a price target of $350, representing an implied over 70% upside from the current price of $204.12.

Valuation Paradox: 7-Year Low P/E, Why Is the Market Discounting?

Nvidia’s current share price of $195.55 corresponds to a projected FY2027 P/E of just 21.5x (BofA estimated EPS $9.09), dropping to 14.7x for FY2028. This is not only Nvidia’s lowest valuation in seven years but also a significant gap compared to its large tech peers.

BofA’s comparative data shows that Apple, Microsoft, Google, Amazon, and Meta are trading at 22x/19x P/E on projected 2027/2028 earnings, about 30-35% higher than Nvidia.

BofA's core point: The AI opportunity and memory cost pressures faced by Nvidia are no fundamentally different from the five companies above, yet Nvidia bears an extra valuation discount.

The logic behind this discount implies the market has already “priced in” a 30-35% drop in Nvidia’s 2027/2028 EPS—a premise BofA finds completely unfounded.

Looking at PEG (P/E to growth ratio), Nvidia’s 2027 PEG is just 0.3x, far lower than Apple’s 2.7x, Microsoft’s 1.0x, and Google’s 1.9x, giving it outstanding relative value.

Meanwhile, BofA believes this deep valuation discount is entirely due to market overreactions to memory costs and competition narratives. The upcoming earnings call will be a key positive catalyst, likely to further clarify to the market Nvidia’s unbreakable moat in products, pricing and supply chain.

Memory Cost: Impact Overestimated, Pricing Power Underestimated

Market concerns about Nvidia’s gross margins focus on the continued rise in high-bandwidth memory (HBM) costs. BofA argues that these worries both overestimate the cost impact and underestimate Nvidia’s pricing power.

The data reveals the true profit margin: Upgrading each rack from the Blackwell to the Vera Rubin architecture, the incremental HBM cost is about $200,000–$300,000; however, the entire rack is expected to be priced higher by $2–3 million (rising from Blackwell’s ~$3–4 million/rack to Vera Rubin’s ~$6–7 million/rack).

The price increase is driven not only by memory, but also by the upgraded Vera CPU, NVLink and Quantum Ethernet networking (which do not require HBM), and a range of software features to accelerate inference and reduce Time to First Token (TTFT) cost.

Nvidia also notes that, compared to Blackwell, Vera Rubin can achieve:

  • About 10x improvement in performance per watt (10x drop in token inference cost);
  • About 3.3x increase in inference capability;
  • Up to 5x increase in training capability.

Furthermore, Nvidia has built a formidable supply chain moat with over $119 billion in pre-purchase commitments in memory, wafers, packaging, and electricity. BofA thus expects Nvidia’s gross margin to remain in the mid-70% range.

ASIC Competition: 700-Fold Growth Has Already Answered the Question

Custom chip (ASIC) threats to Nvidia is another major market concern. BofA responds with these figures:

  • Google TPU launched in 2015;
  • Amazon Trainium launched in 2020;
  • Meta MTIA launched in 2023.

Yet since 2015, Nvidia’s GPU accelerator sales have grown about 700 times. More importantly, per Nvidia’s latest disclosures, its sales to hyperscalers grew 115% year-over-year, about twice the growth rate of overall cloud capital expenditure. This shows Nvidia’s wallet share among hyperscale customers has continued expanding rather than shrinking.

BofA’s logic: ASICs are narrow, function-specific products used only within certain cloud providers, while Nvidia offers a widely available, ecosystem-rich general-purpose platform. Long-term, BofA expects Nvidia will maintain a market share of 65-70%+ of AI compute capex, with the rest split among ASICs, AMD, and others.

AI Market Potential: A $1.7 Trillion Feast by 2030

BofA’s long-term forecasts for the AI data center market provide macro support for Nvidia’s sustained growth. The forecast data:

Global AI data center system market size expects to grow from $273 billion in 2025 to about $1.7 trillion in 2030, a CAGR of 44%;Of this, the AI accelerator (GPU/ASIC/XPU) market is expected to rise from around $197 billion in 2025 to about $1.1 trillion in 2030;The HBM market is expected to grow from $34.5 billion in 2025 to $246.3 billion in 2030, a CAGR of 48%;The AI networking market is expected to grow at a 45% CAGR, reaching $110 billion in 2030;The overall data center system market (AI + non-AI) is expected to grow from $505.6 billion in 2025 to about $2.1 trillion in 2030 (33% CAGR), far exceeding the 9% growth of overall IT expenditure.

For Nvidia’s own EPS trajectory, BofA estimates: FY2027 EPS to hit $9.09, FY2028 at $13.27, with FY2030 EPS potential surpassing $25.

Two Potential Headwinds: Position Concentration and Ecosystem Investment

BofA does not shy away from two real pressures Nvidia is currently facing.

First, institutional position concentration.

Nvidia is the largest constituent of the S&P 500 by market cap, with an active S&P 500 fund position weight at 1.15x its relative market cap and held by 78% of active funds (compared to 81% for its peers). BofA thinks this high position concentration will continue to pressure the share price until the market absorbs additional tech supply.

Second, ecosystem investment scale.

Nvidia’s recent total investment in suppliers and customers is about $65 billion, covering OpenAI ($30 billion), Anthropic (up to $10 billion), Intel ($5 billion), and more. BofA estimates this equals 35% of projected FY2026 free cash flow (about $187 billion), or about 17% of projected FY2027 FCF (about $385 billion).

BofA’s conclusion: This investment is a “yellow warning,” but not enough to materially impair shareholder return—Nvidia still has ample room to increase dividends and buybacks.

 

~~~~~~~~~~~~~~~~~~~~~~~~

The above content is from Wind Chasing Trading Desk.

For more detailed insights, including real-time interpretations and frontline research, please join Wind Chasing Trading Desk·Annual Membership

Risk Warning and DisclaimerThe market involves risks and investments should be made cautiously. This article does not constitute personal investment advice and does not take into account any individual user’s specific investment objectives, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. If you invest based on this, you do so at your own risk.