Goldman Sachs' AI Infrastructure Play: Why Nvidia and Broadcom Are the Core Bets for Long-Term Gains

The artificial intelligence revolution is no longer a distant promise—it's a sprawling infrastructure buildout. As companies race to train and deploy AI models, the demand for specialized hardware and software has created a golden age for select tech giants. Among them, Nvidia (NVDA) and Broadcom (AVGO) stand out as the twin pillars of this transformation, according to Goldman Sachs. Their dominance in AI infrastructure, coupled with strategic positioning and financial resilience, positions them to thrive amid secular tailwinds. Here's why investors should pay close attention.

Nvidia: The Unrivaled Leader in AI's Core Engine

Nvidia's $4 trillion market cap—a milestone achieved in 2025—speaks volumes about its stranglehold on AI's most critical component: GPUs. The company's Q1 2026 revenue surged 69.2% year-over-year to $44.06 billion, fueled by hyperscaler demand for its H100 and H800 chips. These GPUs aren't just hardware; they're the backbone of AI's “training” phase, where models like GPT-4 and DALL-E-3 learn from vast datasets.

But Nvidia's advantage extends beyond silicon. Its CUDA software ecosystem and tools like Omniverse have locked in developers and enterprises, creating a moat against rivals like AMD. Even as competitors tout alternative architectures, Nvidia's rapid innovation cycle—with AI-specific chips like the Grace Hopper Superchip—ensures it stays ahead. Goldman Sachs analyst James Schneider's $185 price target (implying 13% upside) reflects confidence in this trajectory, despite headwinds like export restrictions.

The real long-term play, however, lies in AI inference, where models apply their training to real-world tasks. Goldman estimates this market will balloon as hyperscalers and enterprises shift focus from training to deployment. Here, Nvidia's software stack and partnerships with cloud providers (AWS, Azure) give it an edge in owning the full AI lifecycle.

Broadcom: The Stealth Infrastructure Champion

While Nvidia grabs headlines, Broadcom's $1.31 trillion market cap is underpinned by a different but equally vital role: custom silicon for hyperscalers. The company's chips, designed for data-center efficiency, are critical for cost-effective AI inference. Schneider's Buy rating and $315 price target ($13 upside) hinge on Broadcom's infrastructure software business, which is projected to hit 40% of revenue by 2026.

Broadcom's value lies in its strategic partnerships with tech giants like Microsoft and Google. Its “sole-sourcing” relationships—where it's the exclusive provider for specific projects—ensure recurring revenue and high margins. This contrasts with Nvidia's broader, competitive market, making Broadcom a safer bet for steady cash flows.

Moreover, Broadcom's financial health is bulletproof. Fitch Ratings' “BBB” rating for its senior notes underscores its ability to fund R&D and acquisitions. As AI moves beyond training, Broadcom's focus on scalable, low-power chips and software-defined networking positions it to dominate the “inference economy.”

The Barbell Strategy Validates Both Stocks

Goldman's barbell strategy—investing in both high-performance leaders and cost-effective solutions—explains why Nvidia and Broadcom are core bets. Nvidia's GPUs power the cutting edge of AI research, while Broadcom's chips and software enable mass adoption at scale. This dual approach aligns with the industry's evolution: training is just the start.

Even as AI adoption faces hurdles like regulatory scrutiny and cooling costs, the infrastructure secular tailwind remains intact. Hyperscalers are doubling down on AI, and enterprises are following suit. Goldman expects AI CapEx to grow as companies realize cost savings and new revenue streams—validating both stocks' premium valuations.

Investment Takeaway: A Long-Term Play on AI's Infrastructure

For investors, the case is clear: Nvidia and Broadcom are the gatekeepers of AI's future. Nvidia's software ecosystem and GPU leadership ensure it stays the go-to for innovation, while Broadcom's partnerships and scalable solutions make it indispensable for widespread deployment.

• Buy Nvidia for its monopoly on AI training and software dominance. The $185 price target reflects confidence in its ability to capitalize on inference growth.
• Hold Broadcom for its steady revenue streams and hyperscaler ties. Its $315 target rewards investors who bet on the “barbell” model.

Both stocks face near-term risks—export controls, slowing CapEx, or competitive threats—but their positions in AI's core infrastructure make them decade-long plays. In a world where AI is the new electricity, owning the wires and the generators is the surest path to profit.

The AI era is here, and these two stocks are its cornerstones.