AI's Energy Crisis: Why This Infrastructure Firm Outshines ServiceNow

The AI revolution is straining global energy infrastructure, and the race to meet its voracious power demands is creating both winners and losers. While ServiceNow (NOW) benefits from AI's software-driven efficiency gains, a lesser-known player—nuclear/EPC infrastructure firms—is primed to capitalize on the energy bottlenecks threatening AI's scalability. Here's why investors should look beyond NOW's software-centric model and focus on firms solving the physical challenges of AI adoption.

The AI Energy Crunch: A Crisis in the Making

AI's growth hinges on data centers, which now consume 4.4% of U.S. electricity—a figure projected to hit 12% by 2028. Training and running large AI models requires 24/7 power, driving demand for energy that renewables alone can't reliably supply. For example, generating a single 5-second video with advanced AI can consume 3.4 million joules—enough to power a microwave for an hour.

ServiceNow, a leader in IT infrastructure management software, stands to gain as companies adopt AI tools to optimize workflows. Its Q2 2024 revenue rose 12% to $2.2 billion, with AI-driven offerings like Now Generative AI fueling demand. But here's the catch: AI's energy needs are outpacing even the most aggressive software optimizations.

The Hidden Bottleneck: Why Infrastructure Matters More Than Software

While ServiceNow helps companies manage their IT systems, the real problem lies in the physical infrastructure supporting AI:
1. Grid Capacity Gaps: Data centers require 5–10 gigawatts of power (equivalent to a mid-sized city), but grid upgrades lag. Delays in permitting and supply chain bottlenecks mean projects face 7-year waits for grid connections.
2. Renewables vs. Reliability: Solar and wind can't guarantee 24/7 power, pushing firms toward fossil fuels or nuclear energy. Over 48% of data center electricity comes from carbon-intensive sources.
3. Cooling Costs: Advanced AI chips like NVIDIA's H100 consume 600–750 watts, requiring cooling systems that account for 40% of data centers' energy use.

Meet the Undervalued Infrastructure Play: BWX Technologies (BWXT)

While ServiceNow trades at 48x forward P/E, BWXT, a nuclear infrastructure firm, offers a compelling contrast.

Why BWXT?

• Nuclear Know-How: BWXT designs and manufactures critical nuclear components, including reactor cores and fuel. As tech giants like OpenAI seek nuclear-powered data centers (Stargate's $500B plan includes nuclear energy investments), BWXT's is irreplaceable.
• EPC Exposure: The firm's engineering, procurement, and construction (EPC) services are key to building the 5-gigawatt data centers needed for AI. Its backlog includes projects for DOE labs and utility-scale nuclear plants.
• Valuation: BWXT trades at just 12x forward P/E, with 30% earnings growth expected as AI's energy crisis accelerates demand.

Historically, when BWXT has exceeded earnings expectations, the stock has delivered positive returns. Since 2022, such events resulted in an average return of 2.92%, with a peak gain of 8.64% in July 2025 and a minimum of 3.21% in early 2022. This aligns with the thesis that BWXT's earnings beats are reliable catalysts for growth.

Data-Driven Edge Over ServiceNow

The Risk to ServiceNow: Overlooking the Physical Layer

ServiceNow's software solutions address management of AI systems but ignore the energy and infrastructure constraints that could stall adoption. For example:
- Grid Limitations: A 2025 Deloitte survey found 72% of executives see grid capacity as a top challenge, yet NOW's tools don't solve this.
- Carbon Costs: As regulators push for cleaner energy, firms using fossil-fuel-dependent data centers may face penalties—risks NOW's software can't mitigate.

Investment Thesis: Double Down on Infrastructure

Buy BWXT, Sell NOW on Strength:
- BWXT is undervalued and directly tied to AI's energy needs. Its backlog is growing (up 18% YTD), and nuclear's role in data centers is irrefutable.
- NOW is a “me too” play on AI adoption—its software advantages are replicable, while infrastructure bottlenecks are existential.

Conclusion: The AI Revolution Needs More Than Code

AI's future isn't just about algorithms—it's about the energy to run them. Investors ignoring infrastructure risks are missing the bigger picture. BWXT's role in solving AI's energy crisis offers safer, higher-return opportunities than software alone.

Bottom Line: Buy BWXT at $50/share (30% upside potential) and avoid NOW's overvaluation trap. The real AI winners will be those building the power plants and grids that keep the servers running.

Disclaimer: Past performance does not guarantee future results. Consult a financial advisor before making investment decisions.