Broadcom (AVGO) targets long-term growth on hyperscaler deals and AI chip demand

Broadcom Inc. (NASDAQ: AVGO) is positioning for sustained growth as hyperscale customers increase spending on AI infrastructure and networking silicon. In fiscal 2023, Broadcom reported revenue of $35.8 billion and net income of $14.1 billion, according to its Form 10-K filed with the U.S. Securities and Exchange Commission. The company also closed its VMware acquisition valued at about $69 billion, as announced by Broadcom.

Broadcom is increasingly being valued by investors as an AI infrastructure compounder rather than a traditional diversified chip supplier, as hyperscalers scale training and inference clusters that require both specialized silicon and high-bandwidth networking. The company has positioned its Semiconductor Solutions segment around two durable demand vectors tied to hyperscaler buildouts: (1) multi-year custom silicon programs—often custom accelerators and related ASICs developed in close collaboration with a small set of very large customers—and (2) the switching, routing, and connectivity silicon that increasingly becomes a gating factor as clusters scale out. Broadcom has repeatedly emphasized in investor communications that “AI semiconductor” demand is a key growth driver, while also underscoring the depth and stickiness of its custom-silicon engagement model with large cloud customers.

On the infrastructure side of AI, Broadcom’s Ethernet switching portfolio—marketed for large, scale-out data-center fabrics—anchors its thesis that networking remains a critical attach opportunity regardless of which compute platform ultimately dominates inside the server. The company highlights the Tomahawk family for high-bandwidth switching use cases and positions Jericho for routing and deep-buffer scenarios where traffic management and congestion control matter at scale, aligning with the requirements of modern AI clusters. Broadcom also sells data-center network adapters (NICs) designed to support high-throughput connectivity in these environments.

Financially, Broadcom’s ability to fund aggressive R&D and customer co-design programs is supported by a large and increasingly diversified revenue base—now spanning both semiconductors and enterprise software following the VMware acquisition. For the fiscal year ended Nov. 3, 2024, Broadcom reported net revenue of $51.6 billion, up from $35.8 billion in fiscal 2023, with the increase driven largely by the VMware deal’s contribution in fiscal 2024.

Quarterly reporting has also underscored how both segments are contributing to scale. In the quarter ended Feb. 2, 2025, Broadcom reported total net revenue of $14.92 billion (up 25% year over year), with Semiconductor Solutions revenue of $8.21 billion (up 11% year over year) and Infrastructure Software revenue of $6.70 billion (up 47% year over year). The company also reported GAAP net income of $5.50 billion for the quarter, illustrating the earnings power it can reinvest into next-generation silicon and platform integration.

The market has increasingly linked that scale to AI-driven demand signals. Third-party coverage in March 2026 cited AI semiconductor revenue of $8.2 billion in Q1 FY2026 (up roughly 100% year over year), while Nasdaq market data referenced around the same period placed Broadcom’s market capitalization at roughly $1.5 trillion—context that reflects how investors are discounting sustained hyperscaler spending and long-duration custom silicon roadmaps.

Broadcom’s differentiated positioning—custom AI ASIC co-design plus networking scale—also places it in a distinct competitive lane versus merchant accelerator leaders. While Nvidia and AMD are primarily associated with broadly distributed GPU and CPU/GPU platforms, Broadcom’s exposure is more tightly linked to the “picks-and-shovels” of AI data centers (network fabrics, connectivity, and select custom compute programs) and to a limited set of hyperscaler relationships that can be deepened over multiple product generations.

Broadcom (AVGO) Hyperscaler Partnerships and Custom AI ASIC Strategy: Custom Silicon and AI Networking in Focus

Broadcom’s Hyperscaler Strategy Explained

Broadcom has built a significant position in data center infrastructure by supplying hyperscalers with custom silicon and high-performance networking components used in large-scale AI clusters. In recent earnings materials and investor communications, Broadcom has highlighted “AI semiconductor” demand as a key growth driver, alongside continued investment in custom silicon programs and data center networking that support accelerated computing deployments.

Unlike merchant GPU offerings that are broadly sold across enterprises and cloud providers, custom ASIC programs are typically developed over multiple product generations and require close collaboration on architecture, physical design, verification, packaging, and platform integration. Broadcom has described its custom silicon business as involving deep engagement with a limited number of very large customers, which can lead to multi-year product roadmaps and higher switching costs once platforms are deployed at scale.

Broadcom and several hyperscalers do not always publicly disclose detailed supplier relationships for specific AI accelerators. As a result, references to particular in-house chips should be interpreted as industry context rather than a comprehensive mapping of supplier roles unless confirmed in primary disclosures. Broadcom has, however, consistently positioned itself as a major provider of custom silicon and networking for large cloud customers as AI training and inference clusters expand.

Alongside compute-adjacent silicon, Broadcom’s networking portfolio is designed to address the bandwidth and latency requirements of large AI clusters. The company has announced high-speed Ethernet switching and network interface products intended for scale-out AI fabrics, including the Tomahawk family of switches and high-bandwidth NIC solutions, which Broadcom markets for data center and AI networking use cases.

Financial Performance and AI Revenue Growth

Broadcom has reported rapid growth in “AI semiconductor revenue” in recent quarters and has tied that performance to demand from hyperscale customers building out AI infrastructure.

In its latest reported quarter, Broadcom disclosed specific period financial results across its two reporting segments—Semiconductor Solutions and Infrastructure Software—and provided management commentary on drivers that include AI-related demand, product mix, and supply chain considerations such as foundry and advanced packaging capacity.

Broadcom has also emphasized that its model remains “fabless,” relying on third-party foundries for manufacturing while focusing internal investment on R&D, design, and platform-level engineering. Management has discussed how mix shifts—particularly a higher contribution from AI-related products and systems—can influence margins, alongside external factors such as foundry pricing and advanced packaging availability.

Separately, Broadcom’s infrastructure software business (which includes VMware) adds recurring revenue streams that management has positioned as complementary to the semiconductor cycle.

How AVGO Compares to AI Chip Rivals

Broadcom’s custom silicon and networking exposure in AI infrastructure differs from the market positions of Nvidia and AMD, which are more directly associated with merchant accelerators (GPUs) and CPU/GPU platforms sold broadly across cloud and enterprise customers. Nvidia’s financial disclosures and earnings commentary detail data center growth driven by AI deployments, while AMD has reported accelerating data center revenue tied to EPYC CPUs and Instinct GPU ramps.

Where Broadcom tends to be discussed in the context of hyperscaler customization and networking, Nvidia’s differentiation is frequently attributed to its software ecosystem and platform integration, and AMD’s positioning often emphasizes an open software stack and competitive performance-per-dollar for certain deployments.

Broadcom’s differentiation in AI infrastructure is often framed around two areas:

– Custom silicon engagement model: Multi-year co-development cycles with a small set of very large customers can support longer-duration revenue opportunities, though program timing and volumes can vary by customer build plans.

– Networking at scale: High-bandwidth switching and NIC products are central to scaling AI clusters, and Broadcom is a long-standing supplier of Ethernet switching silicon and connectivity solutions used in data centers.

Analyst Outlook and Key Risks

Analyst and market commentary on Broadcom frequently points to AI-related semiconductor growth, exposure to hyperscaler capex, and the combined semiconductor-plus-software model following the VMware acquisition.

Key risks and sensitivities commonly highlighted in Broadcom’s filings and management commentary include:

– Customer concentration and program timing: A meaningful portion of semiconductor revenue can be tied to a small number of large customers; changes in customer deployment schedules or competitive dynamics could impact results.

– Margin and supply chain factors: Product mix, foundry pricing, advanced packaging availability, and other supply chain constraints can influence gross margin and delivery timelines.

– Competition and platform shifts: Nvidia and AMD continue to invest aggressively in data center platforms, while hyperscalers can adjust their mix of merchant accelerators, internally designed chips, and networking architectures over time.

– Software integration and customer response: The VMware integration introduces execution risk, including product roadmap alignment and customer sentiment around licensing and pricing changes.

As of March 20, 2026 (based on NASDAQ pricing data for AVGO on that date), Broadcom’s market capitalization was approximately $1.5 trillion.

AI Infrastructure Demand

Broadcom has been positioning its semiconductor business around AI infrastructure demand, leaning on two pillars: high-performance Ethernet switching/routing silicon (including its Tomahawk and Jericho families) and a custom-silicon program that management has framed as a growth driver tied to large cloud customers. In its annual report, Broadcom describes its Semiconductor Solutions segment as spanning networking, broadband, wireless, server storage and industrial markets, while also emphasizing that AI-related demand has become a growing driver of networking and custom silicon programs.

On the product side, Broadcom markets Tomahawk as a high-bandwidth Ethernet switching platform designed for scale-out data-center fabrics, while Jericho is positioned for routing and deep-buffer use cases where traffic management and scale matter. Broadcom’s product materials outline how these chips are used to build high-throughput, low-latency fabrics in large clusters.

Broadcom also highlights its ability to serve customers through both merchant silicon and custom designs, which can create longer product cycles and deeper integration into customer roadmaps when engagements are multi-year. The company generally discusses these programs in aggregate in filings and investor materials rather than naming individual hyperscalers or disclosing customer-specific deployments.

AVGO vs. Competition

Broadcom’s positioning in AI data centers is most often discussed through the lens of networking silicon (Ethernet switching/routing) and custom silicon, rather than as a primary seller of merchant AI GPUs. Nvidia remains the dominant merchant accelerator supplier, while AMD has been scaling its data-center GPU presence; Broadcom, by contrast, is more directly exposed to AI cluster buildouts through networking and select custom programs. Broadcom markets its Ethernet switching portfolio for scale-out data-center architectures that support large AI clusters.

Broadcom (AVGO) highlights custom AI silicon demand and VMware-driven software scale as it reports quarterly results

Broadcom (AVGO) reported quarterly results while continuing to position its business around two main themes it says are driving incremental demand: custom silicon programs for large cloud customers and high-speed networking components used to connect AI compute clusters. The company is also leaning on its Infrastructure Software segment following the closing of its VMware acquisition, which Broadcom has described as a key driver of recurring revenue and margin mix.

In its earnings materials, Broadcom said AI-related semiconductor revenue has been growing as hyperscalers build out AI infrastructure and adopt custom accelerators and networking silicon, while management has also emphasized cross-sell opportunities across VMware’s virtualization and management portfolio.

Broadcom’s hyperscaler strategy and custom AI silicon

Broadcom’s custom silicon strategy centers on co-design engagements with large cloud and internet companies that increasingly pursue in-house accelerators and purpose-built chips to manage cost, power, and supply constraints. Broadcom has repeatedly highlighted custom compute (ASIC) and networking attach opportunities as these customers scale AI clusters.

Separately, industry coverage and analyst commentary have described a broader trend of hyperscalers supplementing Nvidia GPUs with internal accelerators and custom ASIC programs, which supports demand for merchant networking silicon (switching, SerDes, optics) that Broadcom sells into AI clusters.

Differentiation from AI chip rivals: Nvidia and AMD

Broadcom’s positioning in AI differs from Nvidia and AMD in that it emphasizes custom silicon engagements and networking components rather than selling a standardized, general-purpose AI GPU platform. Nvidia remains the dominant supplier of AI GPUs and an integrated software stack, while AMD competes with its Instinct accelerator line and broader server CPU portfolio.

Broadcom’s custom ASIC model can lead to deep integration with a customer’s hardware and software environment, but it also tends to concentrate revenue among a smaller number of very large buyers. The company’s AI networking exposure provides an additional lever tied to overall AI cluster scaling, independent of which compute accelerator is chosen.

Analyst sentiment, price targets, and key risks

Wall Street sentiment on Broadcom has been influenced by two factors: (1) whether AI semiconductor demand can remain durable as hyperscalers pace capex, and (2) how quickly Broadcom can scale recurring software revenue and margins following the VMware deal.

Key risks frequently highlighted by investors include customer concentration in custom silicon programs, competitive pressure in AI accelerators and networking, product-cycle execution risk for next-generation silicon, and integration and go-to-market risk associated with VMware. Broader semiconductor cyclicality and the possibility of uneven AI infrastructure spending also remain swing factors for near-term demand and margins.

Broadcom (AVGO) expands custom AI ASIC push with hyperscalers; analysts cite upside while insourcing, supply chain, and regulation remain key risks

Broadcom continues to scale its custom AI accelerator business by partnering with hyperscale customers on application-specific integrated circuits (ASICs), a segment that has become a growing contributor to revenue. In the most recent reported quarter cited by third-party coverage, Broadcom’s AI semiconductor revenue rose sharply year over year, while the stock traded below its late-2025 highs amid broader volatility in AI-related equities.

Broadcom’s exposure to custom silicon demand has also kept analyst focus on contract visibility and execution risks, including customer insourcing, reliance on foundry capacity, and evolving export-control and regulatory regimes.

Broadcom’s AI semiconductor effort

Broadcom has positioned its AI semiconductor effort around long-duration co-design engagements with large technology customers, where it develops custom ASICs for targeted AI workloads and data center deployments. This approach differs from merchant GPU-centric models and is typically tied to multi-year roadmaps and capacity planning.

Recent third-party reporting has pointed to ongoing and potential large custom-ASIC engagements involving major technology platforms and AI developers. For example, TweakTown reported in early 2026 that Broadcom had secured a large custom-ASIC deal and that other prospective customer programs were being discussed in the market.

Broadcom’s hyperscaler strategy is typically described by market observers as:

– Long-term contracts and roadmaps: Custom programs often span multiple product generations, with volumes tied to customer deployment schedules.

– Custom ASIC development: Co-design leveraging Broadcom IP and advanced foundry manufacturing, including production at Taiwan Semiconductor Manufacturing Co. (TSMC).

– Customer diversification: Engagements across cloud, consumer, and AI-native customers, depending on program needs and deployment timelines.

AVGO – competitive differentiation

Broadcom’s AI semiconductor business is frequently discussed alongside Nvidia and AMD, though its role is more concentrated in custom accelerators and networking components tied to hyperscaler deployments rather than broadly distributed merchant AI training GPUs.

Market commentary highlights several differentiators for Broadcom’s custom approach:

– Workload-specific optimization: ASIC designs can be tuned for a defined workload, potentially improving performance-per-watt and total cost of ownership for large deployments.

– Supply-chain planning with customers: Multi-year programs may support coordinated capacity planning across packaging, memory, and foundry supply.

– Reduced dependence on a single merchant platform: Some customers pursue custom silicon to complement, rather than replace, merchant GPU supply.

Competitive positioning table (illustrative; figures depend on company and analyst updates)

Key risk factors

Insourcing and customer concentration

A key risk for Broadcom’s custom silicon business is that large customers may increase internal chip design and shift a larger portion of silicon development in-house. Even in co-design models, program scopes and volumes can change as customers adjust architectures, deployment schedules, or supplier mix.

Supply chain and manufacturing

Broadcom’s advanced-node AI silicon depends on external foundry and packaging ecosystems, which can be sensitive to capacity cycles and geopolitical risk. The company has also detailed supply and manufacturing dependencies in its risk disclosures.

Regulatory and geopolitical exposure

Broadcom operates within a global semiconductor supply chain and customer base, which can be affected by export controls, shifting compliance requirements, and geopolitical developments. Coverage has also pointed to potential sensitivity around customers with cross-border exposure.

Margin compression

Mix shift toward AI accelerators and large custom programs can influence margins versus legacy segments, particularly if costs rise across wafers, advanced packaging, or memory components. Third-party coverage has discussed margin sensitivity alongside AI-driven revenue growth.

Growth outlook

Analyst commentary referenced in third-party reporting suggests expectations for continued AI-driven growth, with forecasts varying by firm and dependent on program timing and customer deployment rates. Investors have also focused on execution against AI-related revenue targets discussed in media coverage.

Broadcom’s scale in networking, custom silicon, and data center connectivity, combined with VMware’s software platform, underpins its medium-term growth narrative. Investors will likely focus on AI-related order visibility, hyperscaler concentration risk, and execution on VMware integration, with primary financial baselines anchored by Broadcom’s reported $35.8 billion in fiscal 2023 revenue and $14.1 billion in net income.

Broadcom’s AI narrative ultimately rests on the idea that hyperscaler AI buildouts are not a one-time hardware wave but a multi-cycle infrastructure expansion—and that the company is positioned in two of the most durable spending categories: custom silicon that can be amortized across massive deployments, and networking silicon that must scale as clusters grow. Broadcom’s communications have consistently framed AI semiconductor demand as a key growth driver, and its product positioning around merchant switching/routing silicon (including Tomahawk and Jericho) reinforces the view that AI’s next bottlenecks often shift from compute to interconnect as models, parameters, and data movement requirements expand.

The company’s reported financial trajectory provides the operating backdrop for that strategy. Broadcom’s FY2024 net revenue of $51.6 billion—up from $35.8 billion in FY2023—highlights how VMware expanded the company’s revenue base and added a large infrastructure software stream that management has positioned as complementary to semiconductor cyclicality. That software scale matters because custom ASIC programs and next-generation networking silicon require sustained, multi-year investment in design, verification, packaging and platform integration, and Broadcom’s enlarged cash-generating footprint can support those commitments through downturns.

Nearer-term results also illustrate both the breadth of demand and the importance of execution. In the quarter ended Feb. 2, 2025, Broadcom delivered $14.92 billion in total revenue (up 25% year over year) and $5.50 billion in GAAP net income, with Infrastructure Software revenue (including VMware) rising 47% year over year to $6.70 billion and Semiconductor Solutions revenue increasing 11% year over year to $8.21 billion. These figures underline a key element of the AVGO thesis: even as AI-linked semiconductors draw headlines, Broadcom’s combined segment mix can provide resiliency while it scales the AI opportunity set across custom compute and networking attach.

At the same time, the long-term outlook is not risk-free, largely because the very characteristics that make Broadcom’s hyperscaler model attractive—large customer concentration, deep co-design, and capacity planning—also create sensitivity to changes in a small number of customer roadmaps. Broadcom’s filings highlight customer concentration and competitive dynamics as recurring risk themes, while the broader AI supply chain introduces additional uncertainty around foundry and advanced packaging availability and pricing. Regulatory and export-control shifts remain a persistent macro variable for global semiconductor supply chains.

Still, market data and third-party commentary suggest investors are continuing to price in durable AI-linked growth. Nasdaq market data around March 2026 cited a market capitalization of roughly $1.5 trillion, while third-party coverage pointed to sharply higher AI semiconductor revenue, including a report citing $8.2 billion in Q1 FY2026 (up about 100% year over year). If that trajectory holds, Broadcom’s opportunity is less about competing head-to-head with merchant GPU vendors and more about monetizing the infrastructure that hyperscalers must buy regardless—high-speed networking fabrics, connectivity and select custom accelerators—while leveraging VMware’s software footprint to broaden customer relationships and stabilize the earnings profile.