Written by Cassian Vance
As the technology sector converges on Taipei for Computex 2026, the artificial intelligence narrative is shifting from a singular focus on general-purpose graphics processing units (GPUs) toward specialized infrastructure. While Nvidia captured the first wave of AI training expenditure, the second wave—dominated by AI inference and deployment at scale—belongs to custom silicon. No company is better positioned to monopolize this transition than Broadcom (AVGO).
Broadcom will report its highly anticipated fiscal Q2 2026 earnings on Wednesday, June 3, following the market close. With the stock hovering near all-time highs and a market capitalization surpassing $2.1 trillion, the stakes are monumental. However, a deep dive into Broadcom’s $73 billion backlog and dual-moat business model reveals why the current valuation may still underestimate the company’s trajectory.
The Custom Silicon Inflection Point
Broadcom’s primary growth engine is its dominance in Application-Specific Integrated Circuits (ASICs). As hyperscalers like Alphabet, Meta Platforms, Anthropic, and OpenAI transition from training foundational models to running massive inference workloads, the economics of computing change. It becomes prohibitively expensive to rely solely on premium, general-purpose GPUs. Instead, these technology giants are partnering with Broadcom to design custom silicon tailored precisely to their specific AI architectures.
The financial impact of this shift is staggering. In fiscal Q1 2026, Broadcom reported total revenue of $19.31 billion, representing a 29.5% year-over-year increase. More importantly, AI semiconductor revenue surged 106% to $8.4 billion.
For the upcoming Q2 report, management has guided for total revenue of approximately $22 billion—a massive 47% sequential acceleration in growth. Within that figure, AI semiconductor revenue is projected to hit $10.7 billion, representing 140% year-over-year growth. This acceleration indicates that the custom silicon ramp is gathering momentum rather than cooling off.
| Metric | Q1 FY2026 Result | Q2 FY2026 Guidance | Implied YoY Growth |
| Total Revenue | $19.31 Billion | ~$22.00 Billion | 47% |
| AI Chip Revenue | $8.40 Billion | ~$10.70 Billion | 140% |
| Free Cash Flow | $8.00 Billion | N/A | N/A |
Perhaps the most astonishing figure surrounding Broadcom is CEO Hock Tan’s assertion that the company has a clear “line of sight” to $100 billion in AI chip revenue by fiscal 2027. Backed by a $73 billion secured backlog, this is not merely an aspiration; it is a contracted pipeline. To put this in perspective, Broadcom’s entire corporate revenue in fiscal 2025 was roughly $51 billion.
The VMware Software Multiplier
If custom hardware provides Broadcom’s growth engine, its software portfolio provides the valuation multiple. The market was initially skeptical when Broadcom acquired VMware for $69 billion in late 2023. However, that acquisition has transformed Broadcom into the ultimate hybrid technology asset.
In Q1, VMware revenue grew 13% year-over-year, contributing to a total contract value exceeding $9.2 billion. While the top-line growth is steady, the cash generation is extraordinary. Broadcom’s overall free cash flow reached $8.0 billion in Q1, representing a remarkable 41% margin. This recurring, high-margin software revenue insulates Broadcom from the cyclicality of the semiconductor industry and justifies its premium valuation. Hardware sets the revenue ceiling; software sets the earnings multiple.
Valuation and Risks
At roughly $430 per share, Broadcom trades at approximately 87 times trailing earnings. By traditional metrics, the stock appears priced for perfection. However, examining the forward price-to-earnings ratio of roughly 40 times reveals a market that is pricing in the massive backlog already secured. Consensus estimates project a 67% increase in Broadcom’s earnings per share in fiscal 2026, followed by another 61% jump in fiscal 2027.
The primary risk to the Broadcom thesis is customer concentration. The company’s top five customers accounted for roughly 50% of revenue in Q1, and its AI chip business relies heavily on six major hyperscalers. If these companies experience a slowdown in cloud growth or push back on pricing, Broadcom’s revenue could swing violently. Furthermore, any execution missteps in integrating VMware or delivering on the aggressive $100 billion AI target could trigger severe multiple compression.
The Verdict: The Real AI Infrastructure Play
Broadcom does not need Nvidia to fail; it simply needs the AI market to mature. As the industry transitions from general-purpose training to application-specific inference, Broadcom is the manufacturer at the center of the shift. With a $73 billion backlog, accelerating sequential growth, and a 41% free cash flow margin, Broadcom remains the premier infrastructure asset for the next phase of the AI revolution.
Broadcom (AVGO) Verdict: BUY
- Current Price:** ~$430.00
- Price Target:** $480.00 (Consensus) / $630.00 (Street High)
- Catalyst:** Q2 FY2026 earnings on June 3, accelerating custom ASIC deployments, and potential upside revisions to the $100B FY2027 target.
The Semiconductor Ecosystem
Broadcom’s upcoming earnings report will set the tone for the broader semiconductor infrastructure space.
Marvell Technology (MRVL) Verdict: BUY
As Broadcom’s primary competitor in the custom ASIC space, Marvell is the “other” major beneficiary of the hyperscaler shift toward custom silicon. With a massive $225 billion backlog tied heavily to Amazon Web Services (AWS) Trainium chips, Marvell offers a slightly lower-multiple entry point into the custom silicon thesis.
Nvidia (NVDA) Verdict: HOLD
Nvidia remains the undisputed king of AI training, but the custom silicon boom led by Broadcom represents the first credible threat to its monopoly margins. While Nvidia’s data center revenue remains robust, the stock’s valuation leaves little room for the inevitable market share normalization as inference workloads shift to custom ASICs.
Taiwan Semiconductor (TSM) Verdict: BUY
Whether hyperscalers buy general-purpose GPUs from Nvidia or custom ASICs from Broadcom, Taiwan Semiconductor manufactures the chips. TSM is the ultimate “picks and shovels” play, capturing the upside of the entire AI silicon ecosystem regardless of which designer wins the architectural battle.
__________________________________________________
Disclaimer: This article is for informational purposes only and does not constitute investment advice. The opinions expressed are those of the author and do not reflect the views of Equities Orbis or its affiliates. Always conduct your own research before making investment decisions.
