TPU challenges GPU, but Bank of America suggests: Buy Nvidia, Broadcom, and AMD.

```

Despite challenges from custom chips like Google TPU that intensify competition in the AI accelerator market, Bank of America still recommends investors to buy NVIDIA, Broadcom, and AMD, given the rapid expansion of the overall market.

On November 27, according to hard AI news, Bank of America stated in its latest research report that although Google is trying to expand its custom chip ecosystem by renting TPU to Meta—which in theory poses a challenge to NVIDIA and AMD—the bank still maintains a strong bullish stance on NVIDIA, Broadcom, and AMD.

Bank analyst Vivek Arya emphasized that custom chips are mainly suitable for companies like Google and Meta with large-scale internal workloads, while GPUs still have irreplaceable advantages in the public cloud and enterprise markets.

Bank of America expects that by 2030, the total addressable market (TAM) of AI data centers will grow fivefold, reaching over $1.2 trillion. Even if NVIDIA's market share normalizes from the current 85% to 75%, its absolute earnings will still see explosive growth.

Bank of America recommends a full buy of the three AI chip giants—NVIDIA, Broadcom, and AMD—believing that current valuations do not fully reflect their long-term profitability.

Model Battle Intensifies: Is TPU Beginning to Encroach on GPU Territory?

The AI hardware sector is currently experiencing a key architectural contest.

The report notes that Google has just released the Gemini 3 model, followed by Anthropic's low-profile launch of Claude Opus 4.5 yesterday, claiming to surpass Google's new model in coding and reasoning tasks.

Bank analysts pointed out that this indicates LLM development is a long-term marathon, and current snapshots may not reflect long-term trends in market share evolution.

Bank of America especially pointed out that hardware-level competition is more critical. Google relies on its decade-honed TPU for model training (both Gemini 2 and 3 were trained 100% on TPU). Crucially, market rumors suggest that Google may rent TPUs to Meta next year and potentially achieve local deployment cooperation with Meta by 2027.

If true, this would directly impact Meta's current GPU suppliers—NVIDIA and AMD. This signifies that custom chips (ASICs) are no longer just self-use tools for cloud vendors and are beginning to have outward facing potential.

Trillion-dollar Track Explodes: Normalization of NVIDIA’s Share Does Not Change Growth Logic

Although TPUs are coming on strong, Bank of America believes this does not change the core logic of “a bigger pie.” The bank gives three reasons: exploding market size, share logic, and still-solid moats:

Bank of America forecasts that the overall TAM of AI data centers will grow about fivefold from $242 billion in 2025 to over $1.2 trillion in 2030. In this context of rapid growth, even if NVIDIA’s market share declines slightly, absolute revenues will still grow significantly.

The bank’s model assumes that as customers develop custom chips to handle diverse workloads, NVIDIA’s market share will gradually normalize from the current 85% to 75%.

Bank of America believes that although custom chips have cost advantages in specific internal workloads (such as the internal needs of Google and Meta), they lack flexibility in the public cloud environment.

Microsoft Azure, AWS, and over 100 emerging cloud service providers require high flexibility, making general-purpose GPUs still the irreplaceable first choice—even Google’s own public cloud (GCP) still uses NVIDIA GPUs.

Additionally, Bank analysts pointed out that general-purpose GPU chips have advantages such as spot supply, multi-cloud portability, a complete software stack, and a larger developer ecosystem. Moreover, tight supply chain conditions and NVIDIA's scale advantage make it unlikely that market share will see dramatic changes in the short term.

Buy All Three Major AI Chip Giants

The report states Bank of America maintains a buy rating for NVIDIA, Broadcom, and AMD.

Although NVIDIA is expected to achieve over 40% sales and EPS growth, its trading multiple is only about 25x. Bank of America believes this undervalues the company and maintains a buy rating with a target price of $275.

Based on data and share forecasts, NVIDIA's EPS will exceed $10 by 2027 and break $20 by 2030.

Broadcom is seen as the biggest beneficiary of the custom chip trend. Bank of America gives it a buy rating and a target price of $400, corresponding to a 2026 P/E of 37x.

The bank expects the company's 2026 AI business revenue to achieve over 100% YoY growth thanks to additional TPU and Anthropic projects.

However, analysts warn that if Google directly licenses more TPUs, it may erode Broadcom's direct market opportunities for developing ASICs for other clients.

The report states AMD, as NVIDIA’s main competitor, remains worth holding. Bank of America gives it a buy rating and a target price of $300, based on a 2027 P/E of 32x.

This also reflects the company's broad growth drivers across CPUs, GPUs, embedded, and gaming sectors.

Despite the risk of cyclical slowdown in the embedded market, there is still significant upside potential for gaining share in AI computing and CPU markets.

As of Wednesday’s closing, the current stock prices of the three companies were $180.33, $397.57, and $214.24, respectively.

This article is from WeChat Official Account "Hard AI". For more cutting-edge AI information, please visit here

Risk Warning and DisclaimerThe market carries risks; investment requires caution. This article does not constitute personal investment advice, nor does it take into account individual users’ specific investment goals, financial situations, or needs. Users should consider whether any opinions, views, or conclusions herein fit their specific circumstances. Investment is at your own risk. ```