For much of the past decade, Nvidia has dominated the artificial intelligence (AI) hardware landscape and become synonymous with high-performance GPUs used in machine learning, generative AI, and data center acceleration. With a staggering market capitalization recently surpassing $4 trillion, Nvidia now commands more than 90% of the AI GPU market largely thanks to its cutting-edge hardware and the software-first approach embodied in its CUDA platform. However, with scale comes new challenges: as growth rates moderate, opportunities emerge for other players.

Two prime contenders are Advanced Micro Devices (AMD) and Broadcom (AVGO), both of which have considerably lower market capitalizations and, more crucially, substantial room to capitalize on the AI revolution sweeping tech and enterprise sectors. This article explores why AMD and Broadcom could deliver stronger returns than Nvidia over the next five years—and what makes them compelling additions to any forward-looking technology portfolio.

Why Nvidia’s Unmatched Scale May Limit Its Returns

Nvidia’s rise has been nothing short of historic. Its graphics processing units (GPUs) underpin the modern AI boom, with the company’s data center revenue surging nine-fold in just two years, to $39.1 billion last quarter. Its GPU architecture and proprietary technologies such as NVLink and CUDA have become industry standards, cementing Nvidia’s leadership.

But as Wall Street often reminds us, the ‘law of large numbers’ suggests outsized gains become harder to achieve as a company’s base expands. Even as AI spending continues to rise globally—IDC projects the worldwide AI software, hardware, and services market will exceed $500 billion by 2027—Nvidia faces intensifying competition for those incremental dollars, particularly in niche AI inference and networking segments where rivals are gaining ground.

AMD: Scaling Up with AI Inference and CPUs

Advanced Micro Devices (AMD) has spent several years expanding beyond its origins in traditional CPUs and graphic cards. Today, it’s the clear No.2 in high-end AI GPUs, but it’s showing remarkable traction in AI inference—the process of deploying pre-trained models for day-to-day use.

According to AMD’s most recent earnings call, major cloud hyperscalers and one of the world’s largest AI model companies are leveraging AMD’s GPUs for inference workloads and enterprise-scale generative AI. While its ROCm software still trails Nvidia’s CUDA in popularity, it is increasingly considered robust enough for production inference, especially as enterprises prioritize versatility and open standards.

The UALink Consortium, a strategic initiative backed by AMD, Intel, Broadcom, and others, is aiming to develop an open interconnect standard to rival Nvidia’s proprietary NVLink. This would enable cloud providers to build clusters with multi-vendor chips, reducing dependency on Nvidia and opening up a multi-billion-dollar addressable market for AMD.

Financially, AMD’s data center revenue was $3.7 billion last quarter, a fraction of Nvidia’s but representing significant upside. Even modest share gains in the GPU or inference space could move the needle, while AMD is also carving out share from Intel in high-performance CPUs—vital as data centers scale up AI computing.

• Market Capitalization: $237B (as of July 2025)
• 52-Week Trading Range: $76.48 – $185.50
• Gross Margin: 45.15%

Looking forward, as AI inference is projected to outpace training in terms of hardware demand by 2026–2028, AMD stands to benefit strongly, especially in cloud and enterprise AI deployments.

Broadcom: The AI Networking and Custom Chip Innovator

Broadcom (AVGO) approaches the AI market from a complementary angle. Rather than competing directly in the GPU space, Broadcom’s strength lies in networking hardware and custom silicon design. Its Ethernet switches and optical interconnects are critical for transporting massive datasets across AI clusters, and as the size of these clusters increases, so does demand for high-speed, low-latency networking equipment.

In the latest quarter, Broadcom’s AI networking revenue jumped 70% as companies scaled up their data center capacity. The company’s flagship ASIC (application-specific integrated circuit) business is booming, helping hyperscale cloud providers like Alphabet (Google), AWS, and Meta develop custom AI chips that often deliver better performance, lower power consumption, and lower total cost of ownership compared to off-the-shelf GPUs.

Management revealed that its three largest custom chip customers plan to deploy up to 1 million chip clusters each by 2027, representing total revenue potential between $60 billion and $90 billion. These massive deployments are expected to drive significant high-margin revenue over the coming years.

• Market Capitalization: $1.3T (as of July 2025)
• 52-Week Trading Range: $128.50 – $281.18
• Gross Margin: 61.72%
• Dividend Yield: 0.84%

Additionally, Broadcom’s acquisition of VMware brings expertise in AI-driven cloud infrastructure. Its Cloud Foundation platform is gaining traction among enterprises seeking to orchestrate AI workloads seamlessly across multi-cloud and hybrid cloud environments—an area expected to expand exponentially as organizations scale up deployment of AI applications.

The Bottom Line: Opportunity in the AI Ecosystem

While Nvidia remains a juggernaut and is well positioned for future growth, its sheer size makes continued outsized returns increasingly challenging. In contrast, AMD and Broadcom, with their smaller but rapidly expanding AI revenue streams, are positioned to seize market share and benefit from the fastest-growing segments of the AI ecosystem, including inference, networking, and custom silicon.

If AMD secures incremental share in AI inference GPUs, or if Broadcom’s custom chips become the backbone of cloud hyperscalers’ AI clusters, each could outpace Nvidia in stock performance in the coming years. Investors interested in the future of AI infrastructure should watch these companies closely as next-generation workloads shift from training to inference and from single-vendor solutions to open, multi-vendor standards.

As the artificial intelligence boom continues, agile and innovative companies like AMD and Broadcom may offer rewarding alternatives—or complements—to Nvidia for both growth and diversification.

Disclosure: The author, Geoffrey Seiler, and The Motley Fool have positions in several stocks mentioned, including AMD, Nvidia, Alphabet, Broadcom, Apple, and Intel. This article is for informational purposes only and does not constitute financial advice. Please review The Motley Fool’s full disclosure policy.