Nvidia’s widening lead over Apple is becoming one of the clearest signs that financial markets are assigning enormous value to the infrastructure powering the artificial intelligence economy.
Nvidia’s market capitalization now stands roughly $1 trillion above Apple’s, following a dramatic two-week divergence in which Nvidia shares gained 18% while Apple stock declined 10%. The move reflects more than a simple rotation between technology stocks.
It highlights how investors increasingly view advanced computing capacity as one of the most strategically valuable resources in the global economy. At the center of the rally is Nvidia’s position in AI accelerators.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Its Blackwell GPU platform has become critical infrastructure for companies building and operating large AI models. Nvidia itself has said it has visibility into more than $1 trillion in cumulative Blackwell and Rubin revenue from 2025 through 2027, illustrating the scale of demand it expects from hyperscalers.
AI clouds, enterprises and other customers. The extraordinary pricing of GPU rental capacity provides another indication of how tight the market has become.
Nebius, one of the emerging AI cloud providers, has reportedly been auctioning access to Blackwell GPU capacity for as much as $10.50 per hour, more than twice June levels. Such pricing suggests that the scarcity of advanced compute is no longer simply a hardware supply-chain issue.
It is becoming a fundamental constraint on companies trying to train models, run inference workloads and deploy increasingly sophisticated AI agents. This creates an unusual economic environment. Cloud computing has been characterized by falling costs as infrastructure scales and competition increases.
AI computing is currently moving in the opposite direction for the most powerful accelerators. Demand is growing so rapidly that available capacity can command premium prices, particularly when organizations need immediate access rather than waiting months for additional infrastructure.
For Nvidia, that dynamic is highly favorable. Every increase in the economic value of AI compute strengthens the argument for continued spending on GPUs, networking equipment and complete AI infrastructure.
Nvidia is no longer simply selling chips; it is supplying a broader computing platform that includes processors, networking, software and complete systems.
Its annual review describes modern data centers as “AI factories” designed to convert energy and computing resources into digital intelligence. Apple represents a very different economic model. Its enormous valuation has historically been supported by consumer hardware, services, brand loyalty and a massive installed base.
The recent 10% decline in its shares shows how quickly investor priorities can change when the market shifts toward companies directly benefiting from AI infrastructure spending. The contrast does not necessarily mean Apple is losing its long-term competitive position.
Apple remains one of the world’s most powerful technology companies and is itself using Nvidia GPUs through cloud infrastructure for some AI workloads. Instead, the divergence suggests that investors are currently rewarding direct exposure to the AI capital-expenditure cycle.
The bigger question is whether these extraordinary GPU prices are sustainable. High rental rates create powerful incentives for cloud providers to build more capacity, while hyperscalers are simultaneously investing billions in their own infrastructure. If supply eventually catches up with demand, compute prices could fall sharply.
For now, the market is sending a powerful message: AI infrastructure has become one of the most valuable assets in technology. Nvidia’s $1 trillion advantage over Apple, combined with soaring Blackwell rental prices, demonstrates how rapidly the economics of computing are changing.
The AI boom is no longer merely about better software models. It is increasingly a race to secure the physical computing capacity required to run them.