Key Points
-
Data centers now account for over 90% of Nvidia’s revenue.
-
Nvidia offers a full AI platform, combining GPUs, networking, systems, and software.
-
Custom ASICs, slowing AI investment, and regulatory scrutiny could pose challenges, yet the stock is positioned for continued growth through next year.
Amid the buzz surrounding tech, AI, and semiconductor equities, many investors seek the next Nvidia (NASDAQ: NVDA). Rather than chase fleeting trends, I consistently return to Nvidia, having observed countless chip cycles. No other company occupies a more central role in today’s AI landscape.
In the first quarter of fiscal 2027 (ended April 26, 2026), Nvidia generated $81.6 billion in revenue, of which $75.2 billion came from data center sales—a 93% year‑over‑year increase that now exceeds 90% of total revenue. This metric outweighs any gaming or automotive recovery, underscoring where the company’s focus lies. Whenever Microsoft, Meta Platforms, Amazon, or Alphabet expand their AI infrastructure, they allocate substantial capital to Nvidia systems. Likewise, sovereign wealth funds and start‑ups describing AI factories typically refer to racks populated with Nvidia hardware and software.
Image source: Getty Images.
Nvidia’s hardware has evolved beyond individual chips. Blackwell GPUs incorporate roughly 208 billion transistors using a proprietary process, with two dies linked by a 10‑terabyte‑per‑second interconnect that functions as a single unit. These GPUs integrate into systems such as GB200 and GB300, which combine Grace CPUs, Blackwell accelerators, and Spectrum networking into cohesive AI racks. Above this foundation lie CUDA, Nvidia AI Enterprise, and a broad suite of tools for training, fine‑tuning, and inference. In essence, Nvidia offers not merely chips but an integrated AI factory‑in‑a‑box.
Why Nvidia heading into 2027?
My confidence in Nvidia through 2027 stems from the convergence of strong demand and firm pricing. AI‑factory compute has emerged as a distinct investable asset class. Nvidia has allied with BlackRock, Blackstone, Goldman Sachs, and other institutions to channel over $500 billion of third‑party capital into AI infrastructure, while providing the underlying platforms.
Lease rates for H100 and B200 GPUs illustrate this trend. Over the past year, H100 pricing rose from roughly $1.70 per GPU‑hour in October 2025 to about $2.35 per GPU‑hour in March 2026, with on‑demand cloud quotes approaching $2.70 per GPU‑hour by June. Blackwell GPUs command higher rates, ranging from $5.30 to $7.05 per GPU‑hour. These figures indicate that Nvidia has not needed to resort to discounting to sustain demand.
The mental side of Nvidia
There is also a psychological dimension that investors often overlook. Nvidia has become the default ticker for AI exposure; retail traders routinely type Nvidia first when seeking AI‑related stocks, and pension funds frequently select it as a straightforward core holding rather than assembling a basket of smaller names. Analyst coverage remains intensive, and forward‑looking price targets still suggest upside from current levels despite the stock’s already substantial rally. While broad popularity can exacerbate pullbacks, it also grants Nvidia reliable access to capital and subjects it to rigorous scrutiny, helping to maintain disciplined execution.
These points do not imply that Nvidia is without risk. Custom ASICs may erode certain workloads, and AI capital expenditures could retreat if macroeconomic conditions deteriorate or key customers curb spending. Global regulators are also scrutinizing the concentration of compute supply. While such factors are relevant, the prevailing fundamentals—robust data‑center revenue growth, end‑to‑end AI systems, firm pricing, and deep market mindshare—continue to align, leading me to view Nvidia as the premier vehicle for capturing the AI infrastructure theme through 2027.
