Broadcom (AVGO): Custom AI Chips for Cloud Giants, Riding the AI Hardware Wave

Broadcom Inc. (NASDAQ: AVGO) has emerged as a pivotal player in the artificial intelligence (AI) revolution, leveraging its expertise in custom chip design to power the infrastructure of the world’s largest cloud providers. As AI workloads skyrocket—driven by generative AI models, machine learning, and data analytics—demand for specialized hardware has surged. While NVIDIA dominates the GPU market for AI training and inference, Broadcom has carved a unique niche by designing custom application-specific integrated circuits (ASICs) tailored to the specific needs of hyperscalers like Alphabet (Google), Meta, and ByteDance. This deep dive explores how Broadcom is riding the AI hardware wave, its partnerships with cloud giants, and the opportunities and risks ahead in 2025.

The AI Hardware Boom and Broadcom’s Strategic Position

The AI hardware market is undergoing explosive growth. Analysts estimate the market for AI chips will expand from $70 billion in 2024 to over $90 billion in 2025, with long-term projections reaching $400 billion by 2030. This surge is fueled by the computational demands of large language models (LLMs) and other AI applications, which require not just raw power but also efficiency and scalability. Broadcom’s custom AI chips—unlike general-purpose GPUs—offer hyperscalers a way to optimize performance for their proprietary workloads, reduce power consumption, and diversify supply chains beyond NVIDIA’s dominance.

Broadcom’s ASIC business thrives on its ability to collaborate closely with cloud giants to co-design chips. For example, Google’s Tensor Processing Units (TPUs), widely used in its AI-driven services like search and cloud computing, are manufactured with Broadcom’s assistance. Similarly, Meta relies on Broadcom for custom silicon to power its AI-driven advertising and content delivery systems. In December 2024, Broadcom CEO Hock Tan revealed that the company now serves five hyperscalers, including two new clients in advanced development—speculated to be OpenAI and Apple—highlighting its expanding footprint.

This strategic positioning is bolstered by Broadcom’s end-to-end portfolio, which includes not just ASICs but also networking solutions like Ethernet switches (e.g., Tomahawk 5) and PCIe interconnects. These components are critical for building the high-speed, low-latency networks that connect AI clusters in data centers, making Broadcom a one-stop shop for AI infrastructure.

The Daily Newsletter for Intellectually Curious Readers

Join over 4 million Americans who start their day with 1440 – your daily digest for unbiased, fact-centric news. From politics to sports, we cover it all by analyzing over 100 sources. Our concise, 5-minute read lands in your inbox each morning at no cost. Experience news without the noise; let 1440 help you make up your own mind. Sign up now and invite your friends and family to be part of the informed.

Join for free today!

Financial Performance and AI-Driven Growth

Broadcom’s financials underscore its success in capitalizing on the AI wave. In fiscal 2024 (ended November 2024), the company reported $12.2 billion in AI-related semiconductor revenue, a staggering 220% increase from $3.8 billion in 2023. This growth propelled AI to account for 35% of its semiconductor revenue, up from less than 5% in 2021. Total revenue for fiscal 2024 reached $51.6 billion, with analysts projecting a 44% year-over-year increase, driven by AI and the VMware acquisition’s enterprise software contributions.

The company’s Q4 2024 earnings beat expectations, with adjusted earnings per share at $1.26, surpassing estimates of $1.20. This performance sent AVGO shares soaring 21% in a single day in December 2024, pushing its market cap past $1 trillion. Looking ahead, Broadcom forecasts first-quarter 2025 revenue above Wall Street estimates, buoyed by sustained AI demand. CEO Hock Tan’s bold prediction of a $60 billion to $90 billion serviceable addressable market (SAM) for AI chips by 2027—based on just three hyperscalers deploying million-accelerator clusters—signals confidence in long-term growth.

Broadcom’s profitability is equally impressive. With gross margins exceeding 75% in its semiconductor segment and a forward P/E ratio of 29.8 (compared to NVIDIA’s 31.03), the company offers a compelling mix of growth and value. However, its $58 billion net debt—accumulated partly from the $69 billion VMware acquisition—raises questions about financial flexibility if growth slows.

Partnerships with Cloud Giants: A Competitive Edge

Broadcom’s success hinges on its deep partnerships with cloud giants. Alphabet, Meta, and ByteDance are confirmed clients, with each leveraging Broadcom’s ASICs to power massive AI clusters. For instance, ByteDance, the parent of TikTok, reportedly uses Broadcom chips to enhance its AI-driven recommendation algorithms, a critical driver of its ad revenue. The addition of two new hyperscalers—potentially OpenAI and Apple—further diversifies its customer base.

OpenAI’s collaboration with Broadcom, alongside TSMC, to build its first in-house AI chip (reported in October 2024) reflects a shift among AI leaders to reduce reliance on NVIDIA’s costly GPUs. Apple’s rumored AI server chip, codenamed Baltra, set for production in 2026, underscores Broadcom’s role in enabling tech giants to integrate AI across their ecosystems. These partnerships are durable, as custom ASICs require years of development and lock in long-term revenue streams for Broadcom.

Beyond ASICs, Broadcom’s networking solutions complement its chip business. The Jericho3-AI fabric and Trident 5-X12 switch, with its neural network capabilities, optimize data center traffic for AI workloads, giving hyperscalers a reason to bundle Broadcom’s offerings. This synergy strengthens its competitive moat against rivals like Marvell, which competes in the custom chip space but lacks Broadcom’s breadth.

Competitive Landscape and Risks

While Broadcom thrives, it faces stiff competition. NVIDIA remains the AI hardware king, with an estimated 90% share of advanced AI chips. Its GPUs excel in training and inference, areas where Broadcom’s ASICs are less dominant. However, as AI shifts toward inference—where custom chips shine—Broadcom could gain ground. Marvell, with clients like AWS and Microsoft, poses another threat, though its $45 billion custom chip market forecast by 2028 suggests room for both players.

Risks loom large, too. A slowdown in hyperscaler spending, as seen with NVIDIA’s recent growth moderation, could dent Broadcom’s trajectory. Analysts predict its revenue growth may drop below 20% CAGR in 2025 from 50% in Q4 2024, potentially spooking investors. Supply chain constraints, particularly TSMC’s limited advanced packaging capacity (e.g., CoWoS), could also delay chip deliveries. Additionally, geopolitical tensions—like U.S. tariffs on Chinese tech—may impact ByteDance-related revenue, though Broadcom’s U.S.-centric hyperscaler base mitigates this risk.

Future Prospects: Riding the Wave or Facing a Crest?

Broadcom’s future looks bright, but not without caveats. Its $60 billion to $90 billion SAM projection assumes hyperscalers like Google, Meta, and ByteDance each deploy million-accelerator clusters by 2027—a plausible but ambitious scenario given current scaling trends. New deals with OpenAI and Apple could push AI revenue beyond $20 billion annually by 2026, especially as inference workloads grow. The VMware integration, meanwhile, promises margin expansion as software complements hardware sales.

Yet, valuation concerns linger. Trading at 33.4x forward earnings—above its 10-year median of 17x—Broadcom’s stock may face pressure if growth falters. A potential AI spending bubble, coupled with macroeconomic headwinds, could test its resilience. Still, its diversified revenue (AI, networking, software) and sticky customer relationships provide a buffer.

Conclusion

Broadcom (AVGO) is riding the AI hardware wave with gusto, leveraging custom chips and networking prowess to serve cloud giants. Its financial strength, strategic partnerships, and innovation in ASICs position it as a leader in the AI infrastructure race. While risks like competition and valuation loom, Broadcom’s ability to capture a $60 billion-plus market by 2027 makes it a stock worth watching. For Substack readers seeking a deep dive, AVGO offers a compelling narrative of growth, execution, and adaptability in the AI-driven future.

DISCLAIMER: None of this is financial advice. This newsletter is strictly educational and is not investment advice or a solicitation to buy or sell any assets or to make any financial decisions. Please be careful and do your own research.