Amazon expands drone delivery to 500 cities while Nvidia pitches GPUs as investable assets

Amazon announced its Prime Air drone delivery service will reach nearly 500 cities and towns in the United States by the end of 2026, expanding its coverage sixfold from the current 11 locations. The service launches soon in Chicago, Syracuse, Cleveland, Atlanta, and Boise metro areas, according to The Verge.

Each Prime Air site covers approximately 175 square miles and delivers packages up to 5 pounds, with most orders arriving around 60 minutes after checkout. Amazon Prime Air VP David Carbon says the service has already delivered "hundreds of thousands of packages to customers" this year.

The expansion comes despite documented incidents of drones crashing into cranes, internet cables, gardens, and an apartment building. Amazon emphasized its "industry-leading Detect-and-Avoid system" alongside the announcement, likely attempting to preempt concerns from new communities.

Nvidia reframes GPUs as long-lived assets

In a separate development that has implications for how AI infrastructure gets financed, Nvidia partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to arrange $500 billion in financing for compute infrastructure. The pitch: GPUs are now an investable asset class.

"This is really the first time that technology chips have become an investable asset class," Nvidia CEO Jensen Huang told CNBC, as reported by The Verge. "These are revenue-generating assets now. They're productive, they're long-lived, they're fungible, they're flexible."

The messaging represents a sharp reversal from Huang's comments at the company's AI conference last year, when he said of previous-generation Hopper chips: "When Blackwell starts shipping in volume, you couldn't give Hoppers away." The contradiction—chips as disposable versus chips as long-lived assets—highlights the tension in Nvidia's positioning.

BlackRock CEO Larry Fink compared the moment to "the very beginning, like what it was when I started in the mortgage-backed securities market in the 1970s," a reference that may not inspire confidence given how that particular financial innovation ended.

The economics currently support Nvidia's claim. Rental prices for older chips have been rising, and Silicon Data projects continued increases through 2028. One cloud provider nearly doubled its Blackwell B200 rental prices during a recent contract renewal.

But the sustainability of this model faces challenges. The AI industry is becoming saturated with data centers, and Chinese open-source models are demonstrating that less compute can achieve competitive results. As former hedge fund manager Mark Rubinstein noted, mortgage-backed securities failed when mortgages were overproduced—a parallel that may apply if GPU capacity outpaces demand.

What changed

For Amazon, the drone expansion represents a significant operational scale-up. Moving from 11 to nearly 500 locations means the company has solved enough of the regulatory, technical, and safety challenges to bet on broader deployment. The 60-minute delivery window positions drones as a premium fulfillment option for small, urgent items.

For Nvidia and its financial partners, the compute-as-asset-class framework attempts to solve a capital problem: AI infrastructure requires massive upfront investment, and traditional tech financing models don't accommodate the scale. By reframing GPUs as productive assets with predictable revenue streams, the structure opens institutional investment channels that weren't previously available.

The risk is that chip generations turn over faster than the financing terms anticipate, leaving investors holding depreciated hardware. Nvidia's own product messaging has historically emphasized rapid obsolescence—a narrative that conflicts with the "long-lived" asset pitch.

Both developments reflect the infrastructure buildout phase of AI deployment: physical delivery networks expanding to support e-commerce scale, and financial structures evolving to fund compute at a pace that outstrips conventional enterprise IT budgets. Whether either model proves durable depends on demand materializing at the scale these investments assume.