Broadcom (AVGO) Q2 2025: AI Semiconductor Growth and VMware Integration Drive Record Results

Broadcom delivered a record $15 billion in Q2 revenue, driven by robust AI semiconductor demand and strong VMware integration, with AI-related sales now accounting for more than half of semiconductor revenue. Leadership signaled confidence in sustaining high double-digit AI growth into 2026, while software momentum from VMware’s transition to subscriptions continues to accelerate recurring revenue and margin expansion.

Summary

  • AI Revenue Acceleration: AI semiconductor revenue reached $4.4 billion, up 46% YoY, powering total revenue to $15 billion, up 20% YoY.
  • VMware Integration Impact: Infrastructure software revenue rose 25% YoY to $6.6 billion, as VMware subscription conversions boosted recurring revenue and margins.
  • Networking and XPU Momentum: AI networking surged over 170% YoY, with new Tomahawk 6 switch gaining strong customer interest for next-gen AI clusters.
  • Margin Dynamics: Consolidated gross margin was 79.4%, but guidance points to sequential pressure from higher custom XPU mix within AI.

Performance Analysis

Broadcom’s Q2 2025 results reflect the company’s strategic pivot toward AI infrastructure and software-driven recurring revenue. Total revenue hit a record $15 billion, up 20% from the prior year, with AI semiconductors the clear engine of growth. Semiconductor revenue reached $8.4 billion, up 17% YoY, as AI-related products contributed more than $4.4 billion, marking nine straight quarters of strong expansion. The AI segment was buoyed by both custom XPUs, custom accelerators for AI workloads, and AI networking, with the latter growing over 170% YoY and representing 40% of AI revenue for the quarter.

Infrastructure software, driven by VMware, delivered $6.6 billion in revenue, up 25% YoY, and now represents 44% of Broadcom’s total revenue. The successful conversion of enterprise customers to the VMware Cloud Foundation (VCF, full-stack private cloud platform) subscription model has accelerated annual recurring revenue and lifted software gross margin to 93%. Consolidated adjusted EBITDA reached $10 billion, up 35% YoY, with free cash flow at $6.4 billion, or 43% of revenue. However, management guided to a sequential gross margin decline in Q3, primarily due to the growing mix of custom XPU shipments within AI, which carry slightly lower margin profiles.

  • AI Segment Drives Top-Line Expansion: AI semiconductor revenue now accounts for over half of segment sales, sustaining a 46% YoY growth rate and forecast to accelerate to 60% in Q3.
  • Non-AI Semis Remain Soft: Non-AI semiconductor revenue was $4 billion, down 5% YoY, with pockets of sequential growth in broadband and enterprise networking, but industrial and wireless lagged.
  • Software Margin Expansion: Infrastructure software operating margin jumped to 76%, up from 60% a year ago, reflecting disciplined VMware integration and subscription model leverage.

Broadcom’s financial profile is increasingly shaped by high-growth AI infrastructure demand and recurring software revenue, but the company faces evolving margin dynamics as custom AI solutions scale.

Executive Commentary

"Q2 semiconductor revenue was $8.4 billion, with growth accelerating to 17% year on year, up from 11% in Q1. And of course, driving this growth was AI semiconductor revenue of over $4.4 billion, which was up 46% year on year and continues the trajectory of nine consecutive quarters of strong growth."

Hawk Tan, President and CEO

"Gross margin was 79.4% of revenue in the quarter, better than we originally guided on product mix... Adjusted EBITDA was 10 billion or 67% of revenue above our guidance of 66%... This year-on-year improvement reflects our disciplined integration of VMware."

Kirsten Spears, Chief Financial Officer

Strategic Positioning

1. AI Infrastructure Leadership and Custom Silicon

Broadcom’s AI semiconductor business, centered on custom XPUs and Ethernet-based networking, is the primary growth lever, with management forecasting 60% YoY growth into Q3 and signaling confidence in sustaining this trajectory into 2026. The company’s focus on custom accelerators, developed with a select group of hyperscale customers, positions it as a critical enabler of both AI training and inference at massive scale. Management highlighted visibility into multi-year customer deployments and noted that inference workloads are now accelerating demand, potentially pulling forward XPU ramp timing.

2. Ethernet Networking as the AI Cluster Backbone

AI networking, built on Ethernet standards, is central to Broadcom’s value proposition for hyperscalers. The Tomahawk switch portfolio, including the new Tomahawk 6 (102.4 Tbps), enables both scale-out and scale-up AI cluster architectures. Management emphasized that Ethernet’s open standard status gives Broadcom a competitive moat, as customers increasingly favor it over proprietary protocols for both performance and ecosystem flexibility.

3. VMware Integration and Software Recurring Revenue

The VMware acquisition is now translating into tangible recurring revenue and margin expansion, as more than 87% of Broadcom’s largest customers have adopted VCF subscriptions. This transition is driving double-digit annual recurring revenue growth and lifting software gross and operating margins, with management expecting continued conversion momentum for at least another year as multi-year contracts renew.

4. Capital Allocation and Balance Sheet Discipline

Free cash flow generation remains robust, with $6.4 billion generated in Q2, supporting $2.8 billion in dividends and $4.2 billion in share repurchases. Management reiterated its priority to reduce leverage toward a 2x debt-to-EBITDA ratio, while retaining flexibility for opportunistic buybacks and potential large-scale M&A if strategic opportunities arise.

5. Margin Management Amid Product Mix Shift

As custom XPU shipments rise within the AI mix, consolidated gross margin faces moderate structural pressure, given XPUs’ lower margin profile relative to networking and software. Management guided for a 130 basis point sequential margin decline in Q3, but emphasized that overall profitability remains strong due to scale and software leverage.

Key Considerations

Broadcom’s Q2 results underscore a business model pivoting toward high-growth AI infrastructure and stable software annuity streams, but investors should weigh both the upside and complexities of this transition.

Key Considerations:

  • AI Growth Sustainability: Management projects sustained 60% AI revenue growth into 2026, citing visibility from hyperscaler roadmaps and rising inference demand, but execution risk remains as deployments scale.
  • Networking Share and Ecosystem Standards: Ethernet’s dominance in AI networking is a competitive advantage, but the landscape could shift if proprietary protocols gain traction or if hyperscalers change architectural preferences.
  • Software Margin Leverage: VMware’s transition to subscriptions is expanding recurring revenue and margins, but the multi-year renewal cycle means conversion tailwinds could moderate in future periods.
  • Non-AI Segment Recovery: Non-AI semiconductors remain weak, with only modest sequential improvement, highlighting Broadcom’s dependence on AI and software for growth.
  • Capital Allocation Flexibility: Strong cash flow supports dividends, buybacks, and debt reduction, but future M&A could alter leverage and capital return priorities.

Risks

Broadcom’s outlook remains exposed to potential volatility in hyperscaler AI investment cycles, competitive threats from alternative networking protocols, and evolving export controls, which management acknowledged as unpredictable. The custom XPU mix shift creates ongoing margin compression risk, and non-AI semiconductor demand recovery is uncertain. VMware subscription conversion tailwinds will eventually normalize as the renewal cycle matures.

Forward Outlook

For Q3 2025, Broadcom guided to:

  • Consolidated revenue of $15.8 billion, up 21% YoY
  • AI semiconductor revenue of $5.1 billion, up 60% YoY
  • Infrastructure software revenue of $6.7 billion, up 16% YoY
  • Adjusted EBITDA margin of at least 66%
  • Gross margin down 130 bps QoQ, due to higher XPU mix

For full-year 2025, management maintained a positive outlook for continued AI and software-driven growth, with confidence in sustaining high AI growth rates into 2026 based on customer visibility and demand for both training and inference workloads. Key factors highlighted include:

  • Continued hyperscaler investment in AI clusters and inference ramp
  • Ongoing VMware subscription conversions and recurring revenue expansion
  • Disciplined capital allocation, with priority on debt reduction and opportunistic buybacks

Takeaways

Broadcom’s Q2 demonstrates the company’s successful transformation into a leading AI infrastructure and software platform, but also surfaces new complexities around product mix, margin management, and customer concentration.

  • AI and Software as Growth Pillars: The company’s results and guidance highlight a business increasingly anchored by AI infrastructure and recurring software revenue, with traditional semiconductor segments contributing less to overall growth.
  • Margin and Capital Allocation Watchpoints: The shift to custom XPUs within AI introduces structural margin pressure, even as scale and software leverage offset some of the impact. Capital return remains robust, but future M&A could shift priorities.
  • Execution and Ecosystem Risks: Sustaining AI growth will require flawless execution with a concentrated customer base, while competitive and regulatory risks remain elevated amid rapid industry change.

Conclusion

Broadcom’s Q2 2025 results validate its leadership in AI infrastructure and software, with strong momentum in both custom silicon and VMware-driven recurring revenue. Management’s confidence in sustaining high double-digit AI growth into 2026 is a clear positive, but investors must monitor margin trends, customer concentration, and the durability of VMware conversion tailwinds as the cycle matures.

Read-Through

Broadcom’s performance reinforces the centrality of AI infrastructure investment among hyperscalers, with Ethernet networking and custom silicon emerging as critical enablers of both training and inference workloads. The company’s success in driving software subscription adoption post-acquisition offers a blueprint for value creation in large-scale software integrations. For the broader semiconductor and infrastructure software sectors, the results underscore the importance of aligning with hyperscaler roadmaps, investing in open standards, and building recurring revenue streams to offset hardware cyclicality and margin volatility.