← Back to RankingsAVGO-2026-Q2

Earnings Call Analysis

AVGO

Q2 2026

Date: 2026-06-03Rank: #46Forward Promise: very_bullish

Broadcom reported record Q2 FY2026 revenue of $22.2 billion, up 48% year on year, driven by AI semiconductor revenue of $10.8 billion, up 143% YoY. Operating margin reached a record 67% and adjusted EBITDA hit 69% of revenue. AI bookings in the quarter exceeded $30 billion against $10.8 billion shipped, with visibility now extending to 2028. The company guided Q3 consolidated revenue to $29.4 billion, up 84% YoY, including $16 billion in AI semiconductor revenue. Management reiterated FY2026 AI semiconductor revenue guidance of $56 billion (~180% growth) and maintained its FY2027 AI semiconductor revenue outlook of over $100 billion.

Bullishness Score

90.00

μ Mean

95.94

σ Uncertainty

1.98

Forward Promise

8.2

Management Tone

Hock Tan was highly assertive and specific throughout both prepared remarks and Q&A, repeatedly using superlatives like 'insatiable' and 'record' while providing granular forward data points. His demeanor was commanding and confident, occasionally correcting analysts directly. The tone remained consistently bullish from prepared remarks through the final question, with no detectable shift in conviction.

Confidence: HIGH — Management volunteered extensive forward-looking specifics including gigawatt deployment timelines, per-customer ramp schedules, and multi-year revenue trajectories without hedging. Tan directly addressed competitive concerns and supply constraints with definitive statements.

Strategic Signals

Broadcom is positioning itself as the primary silicon and networking supplier to the world's leading frontier AI labs, with six core customers now under contract. The strategic emphasis on gigawatts of compute capacity as the unit of measurement signals a shift from chip-level to infrastructure-level thinking. The $35 billion first tranche of the AI XPU platform with Apollo and Blackstone represents a new capital deployment model that could accelerate demand while reducing customer acquisition risk.

The networking business is being leveraged as a critical differentiator and margin offset within the AI portfolio. With Tomahawk 6 shipping for over a year, a 200-terabit switch taping out this quarter, and leadership in co-packaged optics, Broadcom is embedding itself deeper into AI cluster builds. Management's assertion that networking represents ~40% of AI revenue this quarter but will settle around 30% as XPU volumes scale suggests networking growth remains robust even as its share normalizes.

The VMware integration continues to accelerate, with VCF 9.1 adding heterogeneous compute support across AMD, Intel, and NVIDIA platforms. Software revenue guidance of $8.9 billion for Q3, up 31% YoY, indicates the VMware acquisition is compounding positively with AI-driven server demand. The 93% gross margin in software provides critical profit buffer as semiconductor mix dilutes consolidated margins.

Customer diversification is accelerating materially. Beyond the long-standing Google relationship, Broadcom now has contractual commitments with Anthropic (6 GW total), OpenAI (1.3 GW in 2027, 10 GW by 2029), and Meta (3 GW through 2028), plus two additional customers with $6 billion in purchase orders. This broadening reduces single-customer concentration risk while increasing the aggregate demand visibility.

The shift toward rack-scale and infrastructure-level partnerships, rather than pure chip sales, represents a strategic evolution. The AI XPU platform funded by financial partners to deploy 20+ gigawatts signals Broadcom is moving up the value chain from component supplier to compute infrastructure provider, potentially capturing more value per gigawatt over time.

Key Metrics

Total Revenue$22.2 billion+48% YoY

AI Semiconductor Revenue$10.8 billion+143% YoY

Non-AI Semiconductor Revenue$4.2 billion+6% YoY

Infrastructure Software Revenue$7.2 billion+9% YoY

Operating Margin67.3%+200 bps YoY

Adjusted EBITDA Margin69%above 68% guidance

Semiconductor Gross Margin~70%stable

Software Gross Margin93%stable

Free Cash Flow$10.3 billionrecord

AI Bookings (Q2)$30+ billionvs $10.8B shipped

ARR Growth (Software)17%YoY

Cash Position$19.6 billion+$5.4B QoQ

Days of Inventory86 days+18 days QoQ

Guidance

Q3 FY2026 Consolidated Revenue: $29.4 billion, up 84% YoY

Q3 FY2026 AI Semiconductor Revenue: $16 billion, up over 200% YoY

Q3 FY2026 Non-AI Semiconductor Revenue: ~$4.5 billion, up 12% YoY

Q3 FY2026 Infrastructure Software Revenue: ~$8.9 billion, up 31% YoY

Q3 FY2026 Consolidated Gross Margin: ~74%

Q3 FY2026 Operating Margin: ~67%, flat QoQ

Q3 FY2026 Adjusted EBITDA Margin: ~68%

Q3 FY2026 Non-GAAP Tax Rate: ~16%

Q3 FY2026 Diluted Share Count: ~4.94 billion

FY2026 Full-Year AI Semiconductor Revenue: $56 billion, up ~180% from FY25

FY2027 AI Semiconductor Revenue: in excess of $100 billion

Bottom Line

Price Target: N/A

Broadcom is executing what may be the most impressive scale-up in semiconductor history. AI semiconductor revenue of $10.8 billion in Q2, up 143% YoY, with $30 billion in quarterly bookings, provides exceptional demand visibility. The six-customer diversification strategy is working — Anthropic, OpenAI, and Meta are now contributing meaningful gigawatt commitments alongside Google, reducing single-customer risk. The networking franchise (Tomahawk, Jericho, CPO) provides both a margin offset and a competitive moat that makes Broadcom's XPU offerings more compelling as complete cluster solutions. The software segment at 93% gross margin and accelerating growth (31% YoY guided for Q3) provides profit stability even as semiconductor mix compresses consolidated gross margins. Key risks include the structural gross margin decline, Google share erosion, and the unproven AI XPU platform financing model. However, the Q&A revealed visibility to 2028, supply secured through 2029, and management's most confident forward posture in memory. At current growth rates, Broadcom is on track for $56 billion AI semiconductor revenue in FY2026 and $100+ billion in FY2027, making it the primary pure-play beneficiary of the AI infrastructure buildout. The combination of custom ASIC leadership, networking dominance, and software cash flows creates a durable competitive position that justifies a premium valuation.

Forward Promise

Overall: 8.2 · very_bullish

Unhedged confidence backed by 2028 visibility

Q&A provided some of the strongest forward evidence seen in semiconductor earnings: $30B quarterly bookings, visibility to 2028, supply secured through 2029, and confirmed 10 GW for 2027. The near-term demand trajectory is locked in with minimal execution risk visible.

Red Flags

Consolidated gross margin is declining structurally as AI semiconductor mix grows, falling to a guided 74% in Q3 from 77.1% in Q2. While management frames this as a mix effect rather than a structural semiconductor margin decline, the trajectory is clear: as AI semiconductors become a larger portion of revenue, consolidated gross margins will face sustained downward pressure. Management's explicit recommendation to model semiconductor and software margins separately underscores this permanent shift.

The Google relationship, while described as 'very substantial,' carries explicit share risk. Tan acknowledged that 'given the rate of growth of consumption of AI compute... we fully expect that there will be some diversity of sources for them.' This is the oldest and largest customer, and any share loss at Google could have outsized impact given the revenue concentration.

Inventory is building significantly, rising to $4.3 billion with days of inventory increasing from 68 to 86 days quarter over quarter. While management attributes this to securing supply for accelerating second-half AI demand, any demand deceleration could result in inventory write-downs given the custom ASIC nature of much of this product.

The AI XPU platform financed by Apollo and Blackstone introduces new counterparty and execution risk. Building and operating compute infrastructure is fundamentally different from selling chips, and the $35 billion first tranche represents a significant new business model risk that could divert management attention and capital.

Non-AI semiconductor revenue of $4.2 billion, while up 6% YoY, remains a relatively small and cyclical portion of the business. Management's characterization of a 'full cyclical recovery' based on $6 billion in bookings may prove premature if macro conditions deteriorate, and the wireless seasonal decline highlights ongoing volatility in this segment.

Macro Insights

AI Infrastructure Spending: Management described AI compute demand as 'insatiable' with $30 billion in quarterly bookings, indicating that AI capex cycles are accelerating rather than plateauing. The extension of visibility from 2027 to 2028 in just three quarters suggests the buildout is in early-to-mid stages.

Power Infrastructure: Tan highlighted that power infrastructure (power shells, grid connections) is now a key bottleneck driving booking lead times, not just chip supply. This signals sustained demand for power generation and distribution equipment as a secondary AI trade.

Enterprise AI Adoption: Management noted enterprise AI consumption is 'still relatively at an early stage,' but clarified that enterprise demand funnels through frontier model API providers rather than direct XPU purchases, concentrating the compute demand among a handful of providers.

Semiconductor Supply Chain: Broadcom's ability to secure supply through 2029, including wafers and HBM, suggests the worst of semiconductor supply constraints have eased for well-capitalized players, though Tan noted 'getting supply is not just about dropping money.'

Data Center Networking: With Tomahawk 6 shipping for over a year, a 200-terabit switch taping out this quarter, and CPO as the 'de facto standard,' Broadcom's networking franchise is benefiting from the shift to larger AI clusters requiring higher-bandwidth interconnects.

Private Cloud / On-Premise: VMware Cloud Foundation 9.1 with heterogeneous GPU/CPU support is driving accelerated software growth, suggesting enterprises are building private AI clouds rather than relying solely on public cloud, creating a sustained software tailwind.