Amazon’s $60 billion Qualcomm deal lifts an AI strategy’s chip bet nearly 9%

Qualcomm rose 8.71% from September 1 through September 14 after AWS announced a multigeneration AI-chip collaboration that could involve up to $60 billion of purchases, plus $4 billion in Qualcomm stock warrants. Analyst actions supported the move, including RBC lifting its target to $180 from $160 and Piper Sandler upgrading the shares, while Qualcomm targets $15 billion in data-center revenue by 2029. AI-infrastructure exposure also lifted HPE (+9.7%) and Intel (+9.2%), though Amgen fell 12.9% after disappointing pelacarsen trial results raised concerns for Lp(a)-lowering drugs.

Analysis

QCOM’s re-rating should be discounted until the economics of the AWS arrangement are disclosed. The warrant strike below the prevailing share price functions as customer-funded dilution and may imply that reported contract value materially exceeds net economic value after pricing concessions, wafer commitments, and stock compensation. The key underwriting question is whether QCOM can earn data-center gross margins comparable to its handset business; absent evidence of recurring software, networking, or platform attach, custom silicon risks becoming a lower-multiple, customer-concentrated revenue stream.

The competitive read-through is more adverse for AVGO than for NVDA only if AWS shifts meaningful custom-ASIC design wins away from Broadcom; it is not inherently a displacement of GPU demand. AWS can use custom inference silicon to optimize mature workloads while still expanding NVIDIA capacity for training and frontier models. Over the next 1-3 months, order-book disclosures, any revision to QCOM segment reporting, and evidence of foundry capacity reservations matter more than additional analyst target increases; a lack of quantified revenue timing could unwind the narrative premium quickly.

AMGN’s selloff may be directionally justified for the Lp(a) franchise but is likely too broad if the market treats a class-level cardiovascular endpoint as fully invalidated. Different molecule design, patient selection, achieved Lp(a) reduction, and trial duration can produce divergent outcomes, but the next definitive readout is a binary 6-18 month catalyst rather than a near-term earnings driver. Avoid adding outright exposure until management quantifies the probability-weighted commercial impact and pipeline impairment risk; the better near-term expression is relative-value versus diversified large-cap pharma.

Key Decisions for Investors

• Do not chase QCOM after the initial repricing. Establish a 1-3 month long QCOM / short AVGO pair only if QCOM holds above the warrant strike and management provides a revenue ramp, minimum-purchase structure, or gross-margin framework; target 10-15% relative upside if ASIC share gains are verified. Exit if disclosures indicate nonbinding capacity rather than committed purchases, or if incremental data-center margin is below corporate gross margin.
• For a defined-risk bearish expression on the contract economics, buy QCOM 3-6 month put spreads around 10-15% out of the money after any further momentum-driven rally. The payoff is favorable if dilution, implementation timing, or customer-concentration concerns compress the AI premium; risk is a second hyperscaler design win or disclosed backlog that validates a durable platform business.
• Monitor AVGO for a negative read-through rather than shorting immediately. Initiate a tactical short only upon evidence that AWS design allocation is moving from incumbent custom-silicon programs; absent that proof, AVGO retains stronger ASIC execution, networking attach, and customer diversification than QCOM.
• Keep AMGN as a watch-list relative long versus an equal-weight large-cap pharma basket over 6-12 months, not a standalone pre-event position. Enter only after management preserves 2027-28 growth guidance excluding material olpasiran contribution; thesis is falsified by additional Lp(a) endpoint failures, a material pipeline write-down, or guidance that embeds a substantial loss of probability-adjusted franchise value.