Apple has struck a reported $30 billion chip deal with Broadcom, one of its largest custom silicon procurement agreements ever, as it deepens its effort to build its own AI accelerators and network chips rather than rely on external vendors. The scale of the commitment — spread across multiple chip families — signals Apple is serious about bringing more compute infrastructure in-house, reducing its dependence on third-party AI hardware suppliers over time.
Broadcom (AVGO) is the clear direct winner here, with $63.9B in revenue already growing 23.9% YoY and gross margins of 67.8%. Its custom ASIC business has become a critical revenue driver, and a $30B pipeline from a single hyperscaler-grade customer materially de-risks its growth outlook for the next several years. Apple (AAPL), with $416.2B in revenue and 46.9% gross margins, benefits from locking in silicon supply at competitive terms while building long-term leverage over its AI stack.
The headline frames this as Apple 'edging closer to dethroning Nvidia,' but that framing deserves scrutiny. Nvidia (NVDA) generates $215.9B in revenue growing 65.5% YoY with 71.1% gross margins — driven by hyperscaler and enterprise data center GPU demand that Apple's on-device and internal AI chips don't directly compete with. Apple's custom silicon is optimized for its own devices and cloud inference workloads, not the training clusters and external AI infrastructure where Nvidia holds a near-monopoly.
The real trade tension is in Broadcom: the $30B Apple commitment is a multi-year tailwind for AVGO's custom compute segment, but AVGO already trades at a premium and the news may be partially priced in given prior reporting on Apple-Broadcom collaboration. For Nvidia, the risk is more narrative than near-term financial — institutional investors watching Apple's vertical integration could use this as a reason to question NVDA's long-run moat, even if the competitive overlap is limited today.
Key things to watch: how Apple discloses or references this deal in future earnings calls, whether Nvidia provides any color on hyperscaler customer concentration shifts, and whether Broadcom raises forward guidance on the strength of this commitment in its next print.