Stock Markets March 30, 2026

Bank of America Names Its Top AI Compute Picks, Laying Out Targets and Risks

NVIDIA, Broadcom and AMD top the list as infrastructure spending lifts demand for specialized AI hardware

By Leila Farooq NVDA AVGO AMD
Bank of America Names Its Top AI Compute Picks, Laying Out Targets and Risks
NVDA AVGO AMD

Bank of America identified NVIDIA, Broadcom and Advanced Micro Devices as its preferred stocks in the AI compute ecosystem, assigning specific price targets and detailing valuation frameworks and downside risks as AI infrastructure spending supports demand for specialized compute and networking components.

Key Points

  • Bank of America named NVIDIA, Broadcom and AMD as its top AI compute picks and assigned specific price targets tied to 2027 earnings estimates.
  • NVIDIA’s $300 target is based on 28 times its 2027 estimated PE excluding cash and reflects leadership in AI compute and networking markets but faces risks from cyclicality, competition, China restrictions and regulatory scrutiny.
  • Broadcom’s $450 target uses a 26 times 2027 PE and is supported by expected double-digit EPS growth and strong cash generation but is exposed to customer concentration, competition in networking, frequent acquisitions and a $60 billion net debt load.

Bank of America has disclosed its preferred equities in the AI compute segment, placing NVIDIA Corporation at the top of its list, followed by Broadcom Inc and Advanced Micro Devices, Inc. The investment bank supplied explicit price objectives and the valuation metrics underpinning those targets, while also outlining the principal risks that could undermine each company’s outlook.

How the bank set its targets

Each price target is anchored to a multiple of the firm’s calendar year 2027 earnings expectations. For NVIDIA, Bank of America applied a 28 times multiple to its 2027 estimated price-to-earnings ratio excluding cash to reach a $300 objective. Broadcom’s $450 target reflects a 26 times multiple on 2027 estimated earnings, and AMD’s $280 target is based on 27 times its 2027 estimated non-GAAP EPS. The bank also compared these implied valuations to each company’s historical forward PE ranges when assessing relative fairness.

NVIDIA Corporation (NASDAQ: NVDA)

Bank of America’s $300 price objective for NVIDIA uses a 28 times calendar 2027 estimated PE excluding cash, a level that sits within the company’s historical forward PE band of 25 times to 56 times. The bank justified the multiple by pointing to NVIDIA’s leading market share across fast-growing AI compute and networking markets, which has supported its premium valuation.

But the bank also flagged multiple offsetting concerns. These include lumpiness in global AI project spending, cyclicality in the gaming market, and issues related to power access. The firm listed specific downside scenarios: softer consumer-driven gaming demand, elevated competition from other public companies and internal cloud projects, a larger-than-expected impact from China shipment restrictions, volatility in new enterprise and data center sales, the potential for slower capital returns, and increased government scrutiny of NVIDIA’s dominant position in AI chips.

Recent developments noted alongside the firm’s assessment include Cantor Fitzgerald reiterating an Overweight rating with a $300 price target after NVIDIA’s GTC conference, and a financing move by a European AI firm that raised $830 million in debt to acquire thousands of NVIDIA chips for a new data center.

Broadcom Inc (NASDAQ: AVGO)

Bank of America set a $450 objective for Broadcom using a 26 times multiple on 2027 estimated PE. That multiple lies roughly in the middle of Broadcom’s historical forward PE range of 11 times to 41 times and is consistent with a 1 times to 2 times PEG framework that the bank applies for high-growth compute vendors.

The bank supported Broadcom’s valuation on the basis of expected double-digit EPS growth alongside what it described as best-in-class profitability within semiconductors, strong free cash flow generation, and robust returns. The firm also highlighted exposure that could weigh on the stock: semiconductor cycle sensitivity tied to shifts in the AI theme, high revenue concentration with major customers such as Apple and Google with attendant design-out risk, competitive pressure in networking including rising NVIDIA competition, frequent asset acquisitions that raise integration and financial risk, and a substantial $60 billion net debt position.

Operational items cited in the analysis include Broadcom’s win, together with partner Carahsoft, of a five-year, $970 million contract with the Defense Information Systems Agency to consolidate software contracts, and the start of volume shipments of its Tomahawk 6 switch chip series.

Advanced Micro Devices, Inc (NASDAQ: AMD)

Bank of America assigned AMD a $280 price objective based on 27 times its 2027 estimated non-GAAP EPS. That valuation multiple is positioned toward the middle of AMD’s historical forward PE range of 13 times to 58 times. The bank framed the rating around continued AI-driven growth and gains in CPU share, tempered by expectations for slower expansion in cyclically sensitive embedded and game console markets.

Listed downside risks for AMD include execution risk on its first rack-scale MI400 Series product, uncertainty around the timing and size of AI projects in the Middle East, uneven consumer and enterprise spending patterns, reliance on a single outsourced manufacturing partner, and the maturity of the current video game console cycle.

The bank’s write-up referenced AMD’s announced collaboration with Celestica to develop a new rack-scale AI platform, as well as reiterated Buy and Outperform ratings from Aletheia Capital and Wolfe Research, which the bank said reflected analyst confidence in AMD’s AI accelerator development roadmap.

What this means for markets and investors

Bank of America’s selections map directly to pockets of capital spending on AI infrastructure, where demand for specialized compute and networking hardware has been a primary driver. The bank’s valuations and detailed downside casework underscore the sensitivity of these names to both cyclicality in adjacent end markets and to geopolitical and regulatory developments that could affect supply or market access.


Methodology note

The bank’s targets rely on forward-looking multiples applied to calendar year 2027 earnings estimates and compare those multiples to each company’s historical forward PE range. The analysis lists company-specific recent developments and partner actions that inform the near-term operational outlook.

Risks

  • Project lumpiness and cyclicality - AI and gaming market volatility could reduce demand for compute and networking hardware, affecting semiconductor and data center equipment sectors.
  • Geopolitical and regulatory constraints - China shipment restrictions and heightened government scrutiny could materially impact market access and competitive dynamics in the semiconductor industry.
  • Execution and concentration risks - Reliance on a single manufacturing partner, integration risk from frequent acquisitions, and high customer concentration could disrupt supply chains and revenue stability across compute and networking markets.

More from Stock Markets

Carlyle Preparing Defense-Focused Fund to Tap Rising Military Demand Mar 30, 2026 Citigroup Lines Up Lenders for Ecolab’s $4.8 Billion CoolIT Financing Mar 30, 2026 U.S. Equipment Financing Climbs 14.2% Year‑Over‑Year in February, ELFA Survey Shows Mar 30, 2026 Strategy Inc. Halts Week-to-Week Bitcoin Purchases After 13 Consecutive Weeks Mar 30, 2026 Boston Scientific Shares Drop After CHAMPION-AF Data Shows Mixed Results for Watchman Implant Mar 30, 2026