ByteDance, Alibaba and Tencent ordered 35 billion dollars of Nvidia chips. Nvidia can only fill five percent

— RFI 华语 – 法国国际广播电台 (@RFI_Cn) October 20, 2026
Three of China’s biggest technology powerhouses—ByteDance, Alibaba, and Tencent—have placed massive orders with Nvidia for more than three million RTX PRO 5500 workstation-grade graphics cards, totaling upwards of $35 billion, according to exclusive information obtained by Reuters from two people familiar with the matter. The unprecedented buying spree highlights how tightening U.S. commercial export controls are forcing Chinese tech giants to snap up constrained yet compliant hardware to fuel their artificial intelligence ambitions.
Sources familiar with the discussions revealed that Nvidia has privately warned its clients and partners that it cannot guarantee fulfillment of the orders, as Chinese corporate demand has vastly outstripped the American chip designer’s production projections.
The RTX PRO 5500 is a workstation-grade chip designed primarily for high-end professional computing rather than the massive server clusters required to run Nvidia’s elite AI processors. Even so, it remains one of the most powerful computing options legally accessible to Chinese buyers under the current export restriction framework.
Washington’s regulatory architecture explicitly blocks Nvidia from selling its advanced B300 data center chips—built on the Blackwell architecture—to buyers in China, while exports of the H200 (Hopper architecture) are subjected to strict case-by-case reviews and arduous licensing hurdles.
According to insiders, Nvidia has informed partners that actual deliverable volumes may account for a mere 5% to 10% of the total orders placed. Nvidia projects its global manufacturing capacity for the RTX PRO 5500 will cap out at 3 million units in 2027, whereas Chinese procurement requests alone have already surpassed 6 million cards, leaving a massive supply deficit.
Sources noted that the cards carry a unit price of roughly $11,400, driving the potential total value past the $35 billion mark. These previously undisclosed figures underscore an enduring, aggressive appetite among Chinese firms for American hardware despite heavy geopolitical friction.
Nvidia CEO Jensen Huang has previously pointed out that U.S. export curbs reduced the company’s market share in China’s AI chip sector from near-monopoly levels down to practically zero. However, the fervent pursuit of the RTX PRO 5500 demonstrates that substantial market potential persists in China—provided alternative, compliant products can bridge the gap.
Underneath this procurement wave lies a deeper structural issue: as Chinese companies race to build top-tier AI models and government-backed initiatives push technology adoption across multiple industries, U.S. restrictions have exacerbated local chip shortages. While Washington’s policies aim to steer firms toward domestic alternatives—and Beijing actively champions indigenous semiconductors—local hardware makers still face severe hurdles in matching leading-edge American technology, largely due to restrictions on advanced manufacturing equipment.
An Nvidia representative noted that since export limits were introduced in 2022, Chinese workstation and graphics card manufacturers have experienced “unprecedented expansion.” The representative added that “American vendors are caught between multiple dilemmas, including an increasingly outdated U.S. export control framework—which even incorporates products from five years ago—and Chinese countermeasures against U.S. goods.”
ByteDance, Alibaba, and Tencent did not respond to requests for comment. The U.S. Department of Commerce’s Bureau of Industry and Security and China’s Ministry of Industry and Information Technology also declined to comment.
Industry reports point out that the RTX PRO 5500 lags behind the H200 and B300 in technical specs. The latter two are engineered for massive networked deployments to handle the most complex AI training and heavy-duty tasks.
Three sources indicated that Chinese corporations intend to deploy these graphics cards primarily for AI inference workloads—running existing models to generate conversational AI responses. Compared to model training, which devours massive datasets, inference tasks require significantly lower computing power.
Product specifications reviewed by Reuters show that the RTX PRO 5500 closely mirrors the performance metrics of the RTX 6000D, a watered-down variant Nvidia launched mid-last year specifically to comply with U.S. export rules.
Initially, major Chinese tech firms gave the RTX 6000D a cold shoulder, with some outfits outright refusing to purchase it. Buyers generally felt the hardware’s performance failed to justify its steep price tag, while holding out hope that Washington might eventually soften its stance and clear shipments of superior Nvidia chips.
Insiders noted that as export curbs persisted and supply crunches worsened, hopes for policy easing evaporated. Faced with this reality, the RTX PRO 5500 transformed into an attractive alternative. Sources also noted that Nvidia’s supply constraints stem from both surging demand and deliberate internal capacity allocation strategies.
Starting in 2027, Nvidia plans to dedicate the lion’s share of its foundry capacity with Taiwan Semiconductor Manufacturing Co (TSMC) to building its next-generation Rubin architecture platform for AI and supercomputing, rather than the RTX PRO 5500, the people said. TSMC declined to comment. They added that Nvidia has informed clients that product samples are expected to ship in December, with commercial deliveries slated to kick off in the first quarter of 2027.
Industry insiders revealed that the RTX 6000D has since been discontinued, causing its price on the Chinese market to double. Meanwhile, Shenzhen-listed fintech and AI computing service provider Sunline announced in an October 13 filing that it signed an agreement with an unnamed counterparty to acquire 20,000 units of the RTX PRO 5500, while warning investors that component shortages, high market demand, and international technology regulations could disrupt delivery schedules and final quantities.
