Nvidia on Wednesday released new performance data showing that its newest artificial intelligence server can accelerate the deployment of emerging AI models—including two of China’s most widely used mixture-of-experts systems—by as much as ten times.

The announcement marks another attempt by the company to defend its central role in an industry that is quickly moving from model training to mass-scale usage, an area where Nvidia faces a tougher competitive field.

The company’s findings land at a pivotal moment. For years, Nvidia reigned over the training phase of AI development, supplying hardware that powered nearly every major breakthrough across U.S., European, and Chinese labs. But as AI companies now shift their emphasis to serving models to millions of users, the dynamics have changed. The serving layer is far more cost-sensitive, and rivals such as AMD and Cerebras are positioning themselves aggressively with hardware designed to undercut Nvidia’s dominance.

Nvidia’s latest data centers on mixture-of-experts models, an approach that assigns different segments of a prompt to specialized “experts” within the system. The architecture surged into prominence earlier this year after China’s DeepSeek released an open-source model that stunned global researchers with strong performance, modest training demands, and heavy use of Nvidia hardware. DeepSeek’s ability to push its model to the top of global benchmarks without the massive compute spending seen at U.S. labs reshaped conversations about efficiency.

Since then, the mixture-of-experts race has widened. OpenAI integrated the technique into its newer ChatGPT systems. France’s Mistral adopted it in its own high-performance open-source releases. China’s Moonshot AI—one of the country’s fastest-rising AI firms—entered the field in July with its Kimi K2 Thinking model, which quickly climbed global rankings and drew attention for its ability to scale with lower costs.

Nvidia’s goal is to show that even if mixture-of-experts models reduce the need for training runs, its hardware remains essential once the models go into daily use. On Wednesday, the company said its latest AI server, which combines seventy-two of its top chips into a single system connected by high-speed links, improved the inference performance of Moonshot’s Kimi K2 Thinking model by ten times compared with the previous generation of Nvidia servers. The company has seen similar gains on DeepSeek’s models.

Nvidia attributes the leap to two advantages: the ability to cluster a large number of high-end chips into a single machine, and the ultra-fast interconnects that tie them together. These internal networking features remain areas where the company leads its rivals, especially in systems built for inference at scale.

The shift from training to real-world usage has sharpened interest in these performance claims. Serving models to millions of daily users requires hardware that can guarantee low latency, high throughput, and consistent uptime. It also demands energy efficiency, because power costs rise exponentially as usage expands. Nvidia is wagering that its server architecture will help AI companies meet those challenges without needing to redesign or retrain models for alternative hardware.

Meanwhile, AMD is preparing its own competing server built around clusters of its newest AI chips. The company has said its system will debut next year, and analysts expect AMD to push hard on price competitiveness and interoperability with existing enterprise infrastructure. Cerebras is also expanding its footprint with wafer-scale chips that pitch themselves as simpler alternatives for developers managing large workloads.

The broader question borders on how Nvidia will maintain momentum in a market that is no longer defined solely by training. DeepSeek’s rise underscored how quickly the competitive map can change, and China’s AI ecosystem continues to produce open-source models that spread globally within days of release. At the same time, U.S. and European labs are building mixture-of-experts systems that are far cheaper to scale than older transformer-based models.

Amazon.com says it is reassessing its delivery strategy as the company enters a critical negotiation period with the United States Postal Service, following signals that USPS may overhaul its long-standing partnership with the e-commerce giant.

The existing contract, which expires in October 2026, is now at the center of a broader debate over USPS revenue streams, competitive market access, and the future of last-mile logistics in the United States.

In a statement on Thursday, Amazon confirmed ongoing discussions with USPS but said it was surprised to learn of a potential shift in how the Postal Service intends to allocate access to its delivery infrastructure.

“We’ve continued to discuss ways to extend our partnership that would increase our spend with them, and we look forward to hearing more from them soon—with the goal of extending our relationship that started more than 30 years ago,” the company said. “We were surprised to hear they want to run an auction after nearly a year of negotiations, so we still have a lot to work through.”

The Washington Post reported earlier that new Postmaster General David Steiner plans to hold a reverse auction in early 2026. Under the proposal, access to USPS facilities—critical assets that enable wide-reaching, rapid parcel delivery—would go to the highest bidder rather than being granted directly to Amazon, its top customer. The shift would force Amazon to compete with national retail brands and regional logistics providers for the same operational privileges it currently enjoys.

Amazon delivered more than $6 billion in revenue to USPS in 2025, representing about 7.5 percent of the service’s total sales. Losing the contract, analysts say, would be a severe blow to an agency grappling with structural financial challenges, including the long-term collapse of first-class mail volumes, which have dropped more than 80 percent since 1997. USPS recorded a $9.5 billion loss last year, pressured by shrinking mail revenue, the ongoing migration to digital communication, and intensifying competition from private carriers.

USPS did not comment on the report or the current state of negotiations.

Amazon, meanwhile, says it is reviewing all alternatives to ensure uninterrupted delivery performance.

“Given the change of direction and the uncertainty it adds to our delivery network, we’re evaluating all of our options that would ensure we can continue to deliver for our customers,” the company said.

A rapidly shifting parcel market

Amazon has steadily expanded its logistics ecosystem over the past decade, building what many analysts describe as one of the world’s largest private delivery networks. This network includes hundreds of fulfillment centers, last-mile delivery stations, Amazon Air’s growing fleet, and a non-union workforce that gives the company tighter control over labor costs compared to rival carriers.

Amazon Logistics is already a dominant force in U.S. parcel delivery. Last year, it handled 6.3 billion parcels—just shy of USPS’ 6.9 billion—according to figures from the Pitney Bowes Parcel Shipping Index. The data shows Amazon is projected to surpass USPS in parcel volume by 2028, though analysts say that could occur earlier if the USPS-Amazon relationship fractures.

In April, Amazon pledged more than $4 billion to expand its U.S. rural delivery network by the end of next year, a strategic move that may help insulate the company from possible disruptions tied to USPS. Rural routes remain one of USPS’ most vital assets, given its legally mandated universal delivery obligation and extensive physical reach into underserved regions. Amazon’s investment signals that it is preparing to further reduce dependence on outside carriers.

A meeting at the top and political pressure around USPS

Steiner, who took office earlier this year, met virtually with Amazon CEO Andy Jassy on November 14, according to the Post. The agenda of that discussion remains unclear, but the report suggests the relationship between USPS and its most important customer is navigating a pivotal transition.

USPS’ troubles have also attracted attention from U.S. President Donald Trump, who has repeatedly criticized the agency and suggested sweeping changes. In February, Trump said he was considering merging USPS with the Department of Commerce, calling the Postal Service “a tremendous loser for this country.” Democrats pushed back sharply, arguing such a move would violate federal law and undermine the Postal Service’s independent mandate.

Amid these tensions, analysts say the balance of bargaining power is shifting. “USPS needs Amazon a lot more than Amazon needs USPS,” said New York–based ecommerce analyst Juozas Kaziukenas. “Amazon has all the cards in their hands in this case.”

A 30-year partnership at a crossroads

Amazon and USPS have worked together since the mid-1990s, long before Amazon emerged as a global retail powerhouse. USPS played an essential role in the company’s early growth by allowing low-cost last-mile delivery across U.S. households. As Amazon grew into one of the world’s largest logistics operators in its own right, the relationship evolved but remained vital: USPS handles a large share of Amazon’s weekend and rural deliveries, which complement Amazon’s dense urban distribution footprint.

If the reverse auction proposal proceeds, analysts say it would mark one of the most consequential restructurings of USPS’ commercial strategy in years. It would also open the door to new competitors vying for access to USPS infrastructure at a time when parcel volumes, labor costs, and technological pressures are reshaping the industry.

Amazon’s willingness to publicly highlight its surprise at the potential auction signals that negotiations may now be entering a more contentious stage. The company says it remains interested in expanding its business with USPS, but the uncertainty has forced it to consider parallel strategies to safeguard the reliability of its deliveries.