What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
No verified evidence shows ByteDance bought enough Nvidia GPUs for all of China. The headline appears to combine three different developments: reports that ByteDance was China’s biggest Nvidia-chip buyer in 2025, a collective approval for three companies to buy H200 chips, and a proposed overseas cloud cluster that ByteDance was expected to access rather than own. Together, they point to a race to secure AI computing capacity—not a proven national-scale stockpile.
What the reports actually say
The strongest supportable account is that ByteDance has pursued substantial access to Nvidia computing capacity as rules and supply change. But the reports describe different chips, different locations and different stages of procurement. They cannot be added together into a confirmed ByteDance GPU total.
- 2025 purchases: J.P. Morgan’s January 2026 outlook, citing other reporting and research, said ByteDance bought more Nvidia chips than any other Chinese company in 2025. It did not give a definitive unit count or say ByteDance had most of China’s Nvidia inventory. J.P. Morgan’s outlook.
- H200 approvals: Reuters reported that ByteDance, Alibaba and Tencent were approved to purchase more than 400,000 Nvidia H200 chips collectively. The figure was for all three companies, and the report said conditions were still being finalized; approvals did not necessarily mean purchase orders or deliveries. Reuters report via Investing.com.
- Malaysia cloud project: Tom’s Hardware, summarizing a Wall Street Journal report, described a planned cluster of about 500 Nvidia NVL72 GB200 systems—roughly 36,000 GPUs—with an estimated value of $2.5 billion. Malaysia-based Aolani Cloud was reported as the owner and operator; ByteDance was expected to access the computing capacity. That is not the same as ByteDance buying and owning 36,000 GPUs. Tom’s Hardware’s account.
These are not interchangeable figures. One is a reported ranking without a disclosed count; one is a collective purchase approval; and one is a planned cloud deployment involving third-party ownership. The dossier does not establish the final delivered H200 quantity, ByteDance’s total inventory, or whether the full Malaysian cluster was completed and operational by August 18, 2026.
Buying a GPU is not the same as accessing one
GPU headlines often blur several stages: a company may seek approval, place an order, receive a shipment, install hardware, or rent compute from a cloud provider. Each describes a different level of control. In a cloud arrangement, the provider can own and operate the servers while a customer pays to use their capacity. The customer gets computing time, not necessarily title to the GPUs.
#1 Best Overall
- PLEASE NOTE: Exporting an NVIDIA RTX Pro 6000 GPU outside the US requires strict adherence to the U.S. Export Administration Regulations (EAR) and issuance of an export license from the Bureau of Industry and Security (BIS). Compliance and Know Your Customer (KYC) screening may be required as a condition of order acceptance. [NVIDIA Blackwell Streaming Multiprocessor] The new SM features increased processing throughput, and new neural shaders that integrate neural networks inside of programmable shaders | DLSS 4: Multi Frame Generation ensures ultra-smooth frame pacing for lifelike simulations.
- [Double-Flow-Through Design] The RTX PRO 6000 Blackwell features a double-flow-through cooling design, optimizing efficiency and airflow to sustain peak performance under 600W power loads. | [5th Gen Tensor Cores] Deliver up to 3X the performance of the previous generation and support for FP4 precision for faster AI model processing times with reduced memory usage, enabling local fine-tuning of LLMs and generative AI | [4th Gen Ray Tracing Cores] Double the ray-triangle intersection rate of the previous generation to create photoreal, physically accurate scenes and immersive 3D designs with RTX Mega Geometry, which enables up to 100X more ray-traced triangles.
- [PCIe Gen 5] Support for PCIe Gen 5 provides double the bandwidth of PCIe Gen 4, improving data-transfer speeds from CPU memory and unlocking faster performance for data-intensive tasks like AI, data science, and 3D modeling. | [GDDR7 Memory] With 96 GB of GPU memory and 1.8 TB ps bandwidth, it can tackle massive 3D and AI projects, fine-tune AI models locally, explore large-scale VR environments, and drive larger multi-app workflows.
- [DisplayPort 2.1] Achieve unparalleled visual clarity and performance, driving high resolution displays at up to 8K at 240 Hz and 16K at 60 Hz. Increased bandwidth enables seamless multi-monitor setups while HDR and higher color depth support ensures superior color accuracy for precision work, such as video editing, 3D design, and live broadcasting.
- [Universal MIG] Divide a single RTX PRO 6000 Blackwell into multiple isolated instances, each with dedicated resources, allowing for concurrent execution of multiple workloads, optimized GPU utilization, and secure isolation of different applications or users. [WARRANTY] 3 YR Manufacturer's Warranty. Bulk OEM Packaging. Retail Packaging is NOT included.
Location matters too. A cluster in Malaysia is physically outside mainland China, and its operator may be a separate company. That does not by itself answer who can use it, where users or models are located, whether data crosses borders, or how future export-control rules might apply. Nvidia’s position, as reported by Tom’s Hardware, was that it evaluates and clears cloud partners before supplying products. The reported project was described as compliant with rules at the time of that report; that is not a guarantee that every future use or regulatory interpretation will remain unchanged.
Tom’s Hardware also described a possible Indonesian deployment involving more than 7,000 B200 GPUs. It was reported as a possibility, not a completed installation, and should not be counted as hardware ByteDance already controls.
Why ByteDance might want capacity now
There is a plausible business rationale for securing compute early. Large models, video-generation systems and recommendation services can all require substantial computing resources. ByteDance operates products across different markets, including TikTok and Douyin, and access to overseas cloud capacity could support global operations. These are strategic explanations, not a confirmed statement of ByteDance’s internal motive.
Securing capacity can also hedge against uncertainty: U.S. export rules can limit which chips reach China, while Chinese authorities can restrict or condition purchases of foreign chips. A company that can reserve hardware or cloud capacity ahead of a policy change may have more options than one that waits. But overseas hosting brings its own dependencies, including a third-party operator, cross-border data and governance questions, network costs, and the possibility that later rules restrict access.
Different Nvidia chips, different rules
The reports concern products that should not be treated as one generic pool of “Nvidia GPUs.”
Rank #2
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
- H20: A lower-performance accelerator designed for the China market under earlier U.S. export limits. Its availability and customer acceptance have been affected by changing policy.
- H200: A more capable accelerator. The reported 2026 approvals for Chinese customers were conditional, not an unrestricted opening of the market.
- Blackwell systems: The GB200-based systems in the reported Malaysia project belong to a newer product generation. Their possible availability through an overseas cloud operator is a separate matter from direct H200 imports into mainland China.
- Rubin: Nvidia’s subsequent architecture, not covered by the H200 authorization described by the Associated Press.
The Associated Press reported that H200 sales to approved Chinese commercial customers were subject to U.S. supply requirements, third-party review, military-use restrictions and a cap tied to sales to U.S. customers. Its account did not include Blackwell or Rubin in that approval. AP’s explanation of the H200 approval.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Washington and Beijing both shape access
U.S. export controls can determine which products may be shipped to which destinations and under what conditions. Nvidia’s own risk disclosures warn that rules can change quickly and that large GPU clusters may face conditions such as tracking or throttling. The filing describes potential regulatory risks; it does not establish that those measures are active on every ByteDance deployment. Nvidia’s SEC filing.
Beijing’s approach is also not a simple, permanent ban on Nvidia. China wants more domestic control over technology, but firms also have immediate demand for capable chips. In September 2025, Bloomberg reported that Chinese authorities had instructed companies including ByteDance and Alibaba to cancel orders for Nvidia’s RTX Pro 6000D. Later, Reuters reported conditional H200 approvals for ByteDance, Alibaba and Tencent. The shift illustrates competing priorities: limit reliance on a U.S. supplier while preserving access to computing capacity that domestic supply may not yet replace in every use case. Bloomberg’s account of the RTX Pro 6000D order.
Reuters also reported that authorities had discussed requiring H200 buyers to pair imported chips with a specified ratio of domestic chips. Such conditions, if implemented, would make procurement a policy instrument as well as a commercial decision.
Chinese accelerators are improving, and domestic chips may fit policy or supply needs better in some settings. That does not establish that they are equivalent to Nvidia across the board—or that they are irrelevant. Performance depends on the workload, software stack, memory, interconnects and model portability. Training a model, running inference at scale and serving a particular application can involve different trade-offs. Companies may also face the cost of maintaining separate software and hardware systems.
Rank #3
- Professional GPU with Blackwell Architecture
- Blackwell Architecture
- 24GB GDDR7 with PCIe 5.0 & Ray Tracing
- AI Workstation
Why “all of China” cannot be checked from these numbers
To substantiate “enough for all of China,” a report would need a ByteDance-specific quantity, a clear definition of what counts as its capacity, and a credible comparison with the country’s total relevant demand or installed compute. None of the cited figures supplies that denominator.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problems“All of China” could mean every Chinese technology company, the national AI sector, all state-funded data centers, or simply a large share of Nvidia chips available to Chinese buyers. Those are very different claims. A GPU count would also need to distinguish installed hardware from approved orders and rented cloud capacity, and to specify whether it counts H20s, H200s or Blackwell systems. The available reporting does not do that.
J.P. Morgan’s outlook discusses China’s efforts to develop domestic chips and restrictions on foreign chips in some new state-funded inference data centers. It also notes that China has accumulated Nvidia chips through official and illicit channels. Those are elements of an analyst’s assessment, not a comprehensive public inventory against which ByteDance’s holdings can be measured. A ranking as the largest Chinese corporate buyer is not proof of dominance over the country’s total supply.
What would make the story clearer
The next meaningful updates are not another dramatic GPU total, but evidence about status and control: whether conditional H200 approvals turn into orders and deliveries; whether the Malaysian cluster is completed and available to ByteDance; what workloads can run there; and whether U.S. or Chinese rules add new limits on cloud access, ownership or use. Until those details are established, combining purchases, approvals and proposed rentals into a single national-scale number would be misleading.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
