AI Data Center Demand Surge: Why Nvidia and Amazon Are Betting Big on the Future
The race to build out AI infrastructure is intensifying, and two tech giants—Nvidia (NASDAQ: NVDA) and Amazon (NASDAQ: AMZN)—are doubling down on their commitments despite Wall Street’s periodic fits of doubt. Recent reports from April 2025 reaffirm that demand for AI data centers remains robust, driven by soaring compute needs, rising energy requirements, and a strategic focus on long-term growth. Yet challenges like supply chain bottlenecks, geopolitical hurdles, and uneven infrastructure progress complicate the path forward.
Ask Aime: How will Nvidia's and Amazon's AI infrastructure investments shape the future of the tech industry?
Amazon: Full Steam Ahead on AWS Expansion
Amazon’s Vice President of Global Data Centers, Kevin Miller, dismissed Wall Street speculation about paused data center leases as “tea leaf reading,” emphasizing that demand for AI infrastructure is “only going up.” AWS’s Q4 2024 results underscore his confidence: revenue grew 19% year-over-year to $28.8 billion, with operating income surging 61% to $10.6 billion, marking its strongest quarter ever.
Despite supply chain constraints—such as chip shortages and power limitations—Amazon’s capital expenditures hit $26.3 billion in Q4 2024, reflecting relentless investment in datacenter infrastructure. However, delays in executing its ambitious 1,000 MW nuclear-powered datacenter campus (with only the 48 MW first phase expected by 2025) highlight the logistical and regulatory challenges ahead. Analysts note that Amazon’s “near-term buildouts” lag competitors like Microsoft and Google, but its financial firepower ensures it remains a contender in the AI arms race.
Ask Aime: "Will Nvidia and Amazon lead the AI race despite supply chain troubles?"
Nvidia: GPU Demand Fuels Growth, Despite Hurdles
Nvidia’s Senior Director of Corporate Sustainability, Josh Parker, echoed Amazon’s bullish stance, stating there has been “no pullback” in demand for AI data centers. The company’s GPUs, particularly its H100 and H20 series, are the backbone of AI compute, with shipments of accelerators equivalent to over 5 million H100 GPUs by late 2024. This scale drives critical power demand: a single DGX H100 server consumes 10.2 kW, and a 20,480-GPU cluster requires 28.4 MW of power at 80% utilization.
However, geopolitical headwinds loom large. U.S. export restrictions have slowed H20 chip deliveries to Chinese tech giants like Alibaba and Tencent, pushing shipments below the one million-unit target for early 2025. Still, Parker insists that compute demands are rising, not falling, as AI models grow in complexity.
The Energy Equation: 50 Gigawatts by 2027
The AI boom is straining global energy systems. Anthropic co-founder Jack Clark projects that 50 gigawatts of new power capacity—equivalent to 50 nuclear plants—will be required by 2027 to meet AI’s baseload needs. Natural gas is emerging as a critical interim solution, though Amazon and nvidia executives sidestepped questions about the environmental trade-offs of such reliance.
Investment Implications: Growth vs. Execution
Analysts see divergent paths for the two stocks. Nvidia’s dominance in GPU technology and its 58.6% upside potential (per April 2025 estimates) reflect confidence in AI’s sustained growth. Amazon, while benefiting from AWS’s scale, faces execution risks: its 33.1% upside target reflects skepticism about its ability to close the infrastructure gap with rivals.
Supply chain bottlenecks and energy costs pose near-term headwinds. For example, a 20,480-GPU cluster’s annual power bill could hit $20.7 million in the U.S.—a manageable cost—while in Japan, where tariffs are 2.7x higher, similar facilities would face $56 million in energy expenses.
Conclusion: Betting on the Long Game
The numbers are clear: AI data center demand is not slowing. With global critical IT power demand projected to double to 96 GW by 2026, and Nvidia and Amazon at the forefront of this expansion, their stocks remain central to any AI-focused investment thesis.
Yet investors must weigh the risks. Amazon’s delayed infrastructure rollout and Nvidia’s supply chain hurdles are real, but both companies are navigating these challenges with aggressive capital allocation and strategic partnerships. The 50 GW power target by 2027—and the $50 billion in annual energy costs it implies—ensures that AI’s growth will reshape not just tech, but energy markets and geopolitics.
For now, the bet remains on the long-term: those who can build AI infrastructure fastest and cheapest will dominate. Amazon’s financial heft and Nvidia’s technological leadership position them as prime contenders. As Miller put it, “the numbers only go up.” The question is whether the world’s energy grids—and investors’ patience—can keep pace.