As artificial intelligence spending accelerates, veteran tech analyst Dan Niles is keeping his investment thesis simple: if you want core exposure to AI infrastructure, own Nvidia and Broadcom for AI Stocks Boom.
He recently told Fox Business that these two companies sit at the center of the capital-spending wave powering modern AI — from data centers to custom silicon — and together provide broad coverage of the ecosystem.
Nvidia remains the dominant supplier of high-performance GPUs used to train and run large AI models. According to Niles, when hyperscalers ramp AI investment, Nvidia is often the most direct beneficiary because its chips are foundational to AI workloads.
CEO Jensen Huang has positioned the company as a full AI platform provider — spanning chips, networking, and software — making Nvidia a primary lever on AI infrastructure spending.
NVIDIA
Performance-wise, Nvidia has delivered explosive multi-year returns, reflecting how central its hardware has become to generative AI, enterprise AI deployment, and cloud scaling.
Broadcom: Custom silicon and AI Networking Muscle
Broadcom plays a different — but equally critical — role. The company designs custom accelerators and networking components that hyperscale cloud providers rely on to build specialized AI infrastructure.
Niles highlights Broadcom as the go-to partner when companies want tailored silicon optimized for their AI stacks. Its networking leadership also supports the massive data flow AI systems require.
This complementary positioning is why Niles argues that owning both companies effectively captures the majority of infrastructure spend flowing into AI.
Why Niles Believes these Two Cover the AI spend
The thesis centers on capital expenditure trends. Niles notes that expectations for hyperscaler AI spending have surged — effectively doubling from earlier projections. When that spending flows into hardware, it tends to funnel toward:
- GPU compute → Nvidia
- Custom silicon + networking → Broadcom
In his words, investors don’t need to overcomplicate the AI trade: these two companies are positioned where the money is actually being deployed.
For Niles, valuation is at the heart of his AI hardware thesis. He notes that Nvidia currently trades at about 25 times forward earnings (46–47x trailing P/E), only slightly above the broader S&P 500, which sits near 23x.
“That’s a compelling price point” – Niles Said
Adding that Nvidia offers a far more attractive risk-reward profile than software names trading closer to 40x earnings. His broader point: when capital expenditure is rising and valuations remain reasonable, he prefers “selling picks and shovels” — companies that provide the tools for AI — rather than those trying to monetize AI indirectly.
In other words, owning Nvidia and Broadcom is a straightforward way to capture the AI boom without overcomplicating the trade.
Meta Platforms: Great at AI, But Not a Hardware Play
Niles also examined Meta Platforms, which has been highly effective at using AI to drive engagement and revenue. Its Q4 results were impressive:
- Revenue: nearly $60 billion (+24% YoY)
- EPS: $8.88 (+11%)
- Operating margin: 41%
- Ad impressions & price per ad: up 18% and 6%, respectively
Despite the strong numbers, Niles warns that Meta isn’t selling AI infrastructure to others. Unlike Nvidia or Broadcom, Meta lacks a large public-facing language model (like Gemini, Grok, or ChatGPT) and doesn’t have a mature public cloud.
That gap feeds market anxiety, especially with a projected $115–135 billion capex spend. However, Meta’s roughly $50 billion partnership with Nvidia, noted by Reuters, could serve as a key validation that its AI investments are hitting the right technology partnerships.
Dan Niles Comparison of Nvidia, Broadcom & Meta
| Company | Valuation | Capital Expenditure Exposure | AI Relevance / Role |
|---|---|---|---|
| Nvidia (NVDA) | 25x forward earnings / 46–47x trailing P/E | Moderate capex as a supplier; benefits from hyperscaler AI spending | Dominant GPU provider for AI training and inference; critical AI infrastructure vendor |
| Broadcom (AVGO) | Similar to Nvidia; attractive risk-reward profile | Moderate-to-high, tied to custom silicon and networking contracts with hyperscalers | Designs ASICs and networking hardware for AI systems; enables hyperscalers’ custom AI deployments |
| Meta Platforms (META) | Software/advertising comparable: higher P/E (~40x for software peers) | Very high projected capex ($115–135B), including AI/Metaverse investments | Uses AI internally for engagement and monetization; not a public AI infrastructure provider; partners with Nvidia for AI tech |
The Bottom Line: Keep It Simple
Dan Niles sums it up simply: if you want pure AI infrastructure exposure, own Nvidia and Broadcom. They cover the two most critical hardware paths — GPUs for AI training and inference, and tailor-made ASICs for hyperscaler data centers.
Valuation remains reasonable, capital spending is accelerating, and the market’s appetite for AI tools is only growing. For Niles, this is not a complicated trade — it’s about owning the companies building the foundation of AI, rather than chasing every software application hoping to capitalize on it.
“Stop overthinking it. If you own both Nvidia and Broadcom, you’re pretty much covered.”
Dan Niles
