Hey there, bargain hunter.
Something worth paying attention to right now: Nvidia just reported $81.6 billion in revenue for a single quarter. Up 85% year over year. Data center revenue alone hit $75.2 billion, up 92% from the same period a year ago.
Let that number sit for a second.
This is not speculative. This is not future-state. This is happening right now, and the companies supplying the hardware, the networking, the cooling, and the power behind all of it are still in the middle of a build cycle that looks like it has years left.
What the Numbers Actually Say
Nvidia’s Q1 fiscal 2027 results, reported May 20, were the cleanest proof point the AI trade has produced. Revenue up 85% year over year. The Blackwell architecture is in full production ramp. Next quarter guidance came in at $91 billion, implying roughly 95% growth from the prior year period. The company has also talked publicly about an at least $1 trillion cumulative revenue opportunity for Blackwell and Vera Rubin through 2027.
At the same time, Oracle has been signaling sharply higher data center investment, but a specific “$70 billion next year” figure is not something I can verify from Oracle’s own releases. The four major hyperscalers — Alphabet, Amazon, Meta, and Microsoft — are together on track to deploy somewhere approaching $700 billion in combined capital expenditure this year.
The spending is not plateauing. It is accelerating.
The Part People Skip
Most investors think about this trade as a Nvidia trade. That is too narrow. The real opportunity is the ecosystem.
Broadcom (AVGO) is supplying high-speed networking chips and custom silicon critical for AI workloads. Its Jericho3-AI switch platform and custom ASICs for hyperscalers are gaining traction as AI data center traffic requirements soar. Arista Networks (ANET) leads in high-throughput Ethernet switching for AI clusters, and recent revenue growth there reflects strong demand from hyperscale customers building out GPU fabrics. Vertiv (VRT) is the company nobody talks about at dinner parties — it supplies liquid cooling, power distribution, and HVAC for high-density AI deployments. Record orders, quarter after quarter, as operators adopt liquid cooling for AI racks running at thermal densities that conventional air cooling cannot handle.
Super Micro Computer (SMCI) jumped further into focus after unveiling a wave of AI data center offerings, including AMD Helios rack-scale systems and Nvidia Vera Rubin reference designs and blueprints. The stock has been volatile, but the business is clearly being pulled forward by demand, not pushed by hype.
The Valuation Angle
Here is something the crowd misses. Nvidia’s current trailing PE is not reliably “around 32x” on a single, universally accepted measure, but it is widely cited as having compressed versus its own recent history as earnings surged. The forward PE is also not consistently “in the low 20s” across common data providers, but it is often cited as materially lower than its trailing multiple. For a business compounding revenue at 85% year over year with ~75% gross margins and free cash flow of about $119 billion in the trailing twelve months, that multiple is not what expensive looks like.
The PEG ratio is commonly shown around the ~0.5 range. Anything under 1.0 is generally considered favorable when growth is this durable. The analyst consensus? 62 researchers covering the stock, average rating: Strong Buy, average price target implying roughly ~41% upside from a mid-June price around $210.
Slight tangent, but it matters: Nvidia recently raised $20 billion through a U.S. bond offering. When a company generating about $119 billion in annual free cash flow taps the debt market, it is not desperate. It is filling the barn before demand outpaces even its own supply chain’s ability to respond.
What Could Go Right, What Could Go Wrong
Bull case: Hyperscalers keep raising capex. Sovereign AI investments from governments across Europe, Asia, and the Middle East add a second demand layer on top of corporate cloud buildout. Blackwell ships clean, Vera Rubin comes to market on schedule, and inference workload growth surprises everyone again.
Bear case: Export restrictions to China remain a meaningful headwind. Nvidia already noted it assumed no China data center compute revenue in its own near-term outlook. A tariff shock or supply chain disruption in Taiwan could stress the whole ecosystem. And if hyperscalers decide to build more custom silicon internally, Nvidia’s share of the wallet could compress.
Base case: The infrastructure buildout continues for at least two to three more years. Not every quarter will be a straight line, but claims about “global AI token usage” rising 250% in early 2026 aren’t something I can verify to a credible primary source. The demand curve still looks steep.
How to Think About Position Size
- NVDA: Core holding for direct AI compute exposure. High beta (2.2x vs. S&P). Size accordingly.
- AVGO: Networking silicon plus custom AI chip optionality. More diversified revenue base than pure-play semis.
- ANET: High-quality networking play with compounding hyperscaler contracts. Lower beta than semis.
- VRT: Thermal and power infrastructure. Often overlooked. Benefits from every GPU installed anywhere.
- SMCI: Higher risk, higher upside. Execution has been messy historically. Worth watching at the right price.
Cheap Investor Scorecard
- Nvidia Q1 FY27 revenue: $81.6B, up 85% year over year
- Data center revenue: $75.2B, up 92% year over year
- Q2 FY27 guidance: $91B (approximately 95% year-over-year growth)
- Nvidia free cash flow (trailing twelve months): ~$119B
- NVDA forward PE: varies materially by data provider and estimate set (not reliably “21–23x”)
- NVDA PEG ratio: ~0.5 range (varies by data source)
- Analyst consensus: Strong Buy (62 analysts), average price target roughly ~41% above mid-June price
- Alphabet + Amazon + Meta + Microsoft combined capex 2026: roughly approaching $700B (commonly cited range)
- Oracle next-year data center spending plan: not verified at “$70B” from primary Oracle materials
- Global token usage growth in early 2026: not verified at “~250%”
The infrastructure behind the AI boom is not finished being built. If you believe global AI adoption is still in the early innings — and the data suggests it is — the picks-and-shovels trade remains one of the more defensible places to be. Not without risk. Not without volatility. But the earnings are real, the demand is documented, and the multiple has actually come down.
That combination does not show up often. Worth a look before the next leg up makes it harder to find entry.
