Qualcomm 在本文所述时点宣布将在今年晚些时候向一个未具名的「大型超级企业级云端客户（large hyperscaler）」交付客制化 ASIC，直接把公司推向与 Broadcom 对抗的 AI 资料中心晶片竞争；同时指出其自有设计依旧按既定节奏推进。这一进场可被视为对智慧型手机事业压力的回应，因为 AI 需求正在把晶片供应由行动端转向资料中心运算。若采纳 IDC 提到的情境，智慧型手机市场（含 13% 的衰退讯号）正加剧结构性转向。

在诠释脉络中可见资本支出正进入前所未见的高增速区间。Meta 将全年资本支出上调至高达1450亿美元（145 billion），高于分析师预估；Amazon 的预算超过2000亿美元（>200 billion），其中多数用于资料中心。Microsoft 报告 Azure 收入增速为39%，并表示扩充运算能力仍吃紧。Google 云端季度营收达到200亿美元（$20 billion），且资本支出可升至高达1900亿美元（$190 billion）。这些数据与 AI 相关的晶片产业今年销售有望首次突破1兆美元（$1 trillion）相互印证。

消息发表后，盘后股价几乎上涨14%，但 AI 硬体赛局本已高度拥挤。Meta、Amazon、Microsoft、Google 等巨头正加速建立内部晶片架构以降低部件依赖，Nvidia 仍是核心参考点；过去被看衰的 Intel 也透过通用 CPU 重新回到 AI 系统供应链。Cristiano Amon 表示 Qualcomm 将采取「all of the above」策略，以客制化晶片、处理器与网路技术全方位投入，争取获得这些 hyperscaler 的设计采购。

Qualcomm announced late in the cycle that it will ship a custom ASIC later this year to an undisclosed large hyperscaler, putting it in direct competition with Broadcom in AI data-center silicon. It also said its own designs remain on track. The move is interpreted as a response to pressure in the smartphone business as AI demand is pulling supply toward data-center compute. IDC’s cited view that smartphone unit economics are under pressure (including a 13% decline) reinforces that shift.

The context shows an unprecedented capex wave. Meta lifted its full-year spending outlook to as much as $145 billion, above analyst expectations. Amazon is budgeting over $200 billion, mostly for data centers. Microsoft reported Azure revenue growth up 39% while saying it is still struggling to deploy enough compute fast enough. Google said cloud unit sales reached $20 billion in the quarter and capex could rise to as much as $190 billion. Together these data points are consistent with forecasts of AI-related chip industry revenue topping $1 trillion this year.

Investors reacted positively, pushing Qualcomm shares up almost 14% after hours, but the AI hardware contest is already crowded. Large hyperscalers are increasingly building internal chip programs to reduce dependency, and Nvidia remains the central reference point for AI systems. Intel’s CPUs are also being redeployed as an AI compute option, reversing some prior irrelevance. CEO Cristiano Amon said Qualcomm will do “all of the above”—custom chips, processors, and networking—to win these hyperscalers’ attention.