Thursday, June 25, 2026
EN·DarkSubscribe
AI Infrastructure · News & Analysis
HomeChips & HardwareReport
Chips & Hardware · Report

Broadcom announced a target of over $100 billion in AI semiconductor revenue by 2028, signaling commitment to custom silicon and ASICs for hyperscaler workloads.

Major semiconductor vendor cementing AI as core business; creates competitive pressure on all chipmakers to develop custom solutions for inference and training.
Trade pressSlicast · June 25, 2026 · US · Source: Google News
importance 85

Broadcom has reported $8.4 billion in AI revenue for a single quarter, representing 106% year-over-year growth. According to the company's guidance, this marks only the beginning of an aggressive expansion trajectory. The semiconductor giant is targeting over $100 billion in annual AI semiconductor revenue by fiscal year 2027, with some analysts projecting figures could exceed $150 billion by FY2028.

The company's booking pipeline underscores the strength of this demand. Bookings have exceeded $30 billion in recent quarters, and CEO Hock Tan has highlighted long-term contracts that extend visibility well into 2028. The quarterly progression illustrates the velocity of growth: Q1 FY2026 delivered $8.4 billion in AI revenue, Q2 guidance came in at $10.7 billion, and Q3 guidance jumped to $16 billion. For the full fiscal year 2026, Broadcom expects approximately $56 billion in AI revenue, representing a 180% year-over-year increase.

Broadcom's recent June 2026 earnings report showed strong growth across the board, though Q3 AI revenue guidance landed slightly below some analyst expectations, creating some stock volatility.

The company has carved out a leadership position by building custom AI accelerators—known as XPUs or ASICs—designed specifically for hyperscale cloud providers. Its customer roster includes Google, Meta, Anthropic, and OpenAI. The OpenAI partnership is particularly significant, with the two companies collaborating on custom chips with production slated to begin in 2026.

Beyond accelerators, Broadcom's networking solutions form a second critical pillar. Technologies including Jericho and Memory fabric serve as the connective tissue inside AI data centers, handling the data flows that make large-scale training and inference possible.

Key risks to the timeline include production hiccups at TSMC, where Broadcom's advanced chips are fabricated, and competition from in-house chip development efforts at Amazon, Google, and Microsoft—initiatives that could eventually reduce their dependence on third-party suppliers like Broadcom.

Read the original
Broadcom announced a target of over $100… · Slicast