This is the supporting evidence and relationship deep dive. For the integrated company view, read the full ByteDance thesis →
COMPANY ECOSYSTEM EVIDENCE • UPDATED 30 SEP 2026
ByteDance infrastructure ecosystem: evidence and relationships
The useful way to model ByteDance is not simply as a consumer internet company that buys GPUs. Its consumer AI demand, Seed research and Volcano Engine cloud business pull it downward into scheduling, heterogeneous accelerators, proprietary networking, storage, custom CPUs, data-centre campuses and grid-scale power infrastructure.
Working classification: Demand-to-infrastructure hyperscaler. ByteDance starts with enormous application and token demand, then engineers downward through the stack to make that demand economically serviceable. That makes it a strong real-world test of the site's central thesis: heterogeneous compute islands can become more useful when networking, software, scheduling, data systems and physical infrastructure are designed as one system.
The ByteDance machine
Scroll horizontally to explore the full diagram →
China AI Infrastructure · chinaaievolution.com · See accompanying evidence and review date.
What the research changed
1 • It is not merely buying acceleratorsByteDance's Seed infrastructure team explicitly owns distributed training, high-performance inference and heterogeneous-hardware compilation. COMET is deployed in production clusters at tens-of-thousands-of-GPU scale.
2 • Networking is an owned optimisation layerByteDance developed veRoCE for RDMA and EthLink for Ethernet scale-up. Company-reported testing says veRoCE improved LLM training speed 11.2% and All-to-All throughput 48.4%.
3 • Physical infrastructure is now strategicWuhu and Wuwei campuses include dedicated 220kV substations. The first Wuhu phase was described by China Energy Engineering as ByteDance's first East China data-centre campus, with total investment expected around RMB8bn.
4 • CPU is becoming part of the AI stackReuters reports ByteDance is developing Arm and RISC-V custom CPUs as inference and agent workloads increase CPU demand; Arm separately identified ByteDance as a customer of its AI data-centre CPUs.
5 • Domestic GPU diversification is realReuters says Huawei is ByteDance's largest domestic accelerator supplier, followed by Cambricon and Iluvatar CoreX. Iluvatar's 2026 allocation was reportedly doubled to 100,000 GPUs as constraints intensified.
6 • Geography is becoming part of orchestrationInner Mongolia's development commission disclosed talks with ByteDance on compute monitoring, trading and converting regional compute advantages into token output. This is emerging policy/market evidence, not proof of production-scale dynamic scheduling.
The accelerator ecosystem
EDGE
STATUS
WHAT IS ACTUALLY EVIDENCED
WHY IT MATTERS
ByteDance × Huawei
CORROBORATED REPORTING
Reuters sources in September described Huawei as ByteDance's largest domestic AI-chip supplier.
Strongest domestic compute edge.
ByteDance × Cambricon
CORROBORATED REPORTING
Reuters describes Cambricon as the second-largest domestic supplier.
Confirms multi-vendor compute.
ByteDance × Iluvatar CoreX
REPORTED
June talks targeted at least 50k inference chips; September reporting said allocation doubled to 100k units.
Inference demand can pull smaller domestic vendors into hyperscale production.
ByteDance × Kunlunxin
CONSIDERED / NOT CONFIRMED
Reuters reported ByteDance was considering Baidu Kunlunxin in June. No sufficient evidence found here of completed production deployment.
Keep as research gap.
ByteDance × Qualcomm
REPORTED
Reuters, citing Bloomberg, reported an AI-ASIC deal involving millions of chips for data-centre/agent workloads; neither company commented.
Potential custom-silicon route beyond GPUs.
ByteDance × Arm
CONFIRMED BY ARM
Arm CEO identified ByteDance as a customer of its AI data-centre CPUs; Reuters separately reports custom Arm/RISC-V CPU development.
Inference shifts system bottlenecks toward CPU as well as GPU.
HBM is the stress test. The September HBM shortage is not peripheral to ByteDance. Reuters reported price increases across Huawei, Cambricon and Iluvatar and said Iluvatar diverted internally earmarked GPUs to ByteDance. This is precisely where the thesis has to become economic: can ByteDance's scheduler, network, model architecture and heterogeneous sourcing offset enough of the higher component cost?
Networking: ByteDance is building around the chip
veRoCE
ByteDance's high-performance RDMA protocol adds multipath/out-of-order optimisation, SACK retransmission and multipath congestion control. Volcano Engine says IANA assigned it UDP port 4794, helping multi-vendor hardware recognise the protocol.
+11.2%
ByteDance-reported LLM training-speed improvement in testing.
EthLink
ByteDance has published an Ethernet-optimised GPU scale-up interconnect design aimed at low-latency, high-bandwidth communication inside AI racks, complementing scale-out networking.
Scale-up + scale-out
The infrastructure strategy is moving from generic network plumbing toward AI-specific fabric design.
Control plane and useful compute
This is where ByteDance becomes particularly important to our thesis. Its compute-infrastructure roles describe unified scheduling across containers, VMs, online/offline workloads and CPU/GPU resources, including intelligent scheduling across CPU, GPU, memory, network and even power across global data centres. Volcano Engine's public products expose the same direction externally through heterogeneous GPU support, K8s scheduling and unified resource pools.
Interpretation — not a company claim: ByteDance is one of the clearest Chinese examples of heterogeneous islands connected by an increasingly intelligent control plane. It does not need Huawei, Cambricon, Iluvatar, Nvidia, Arm and future custom silicon to be interchangeable. It needs the software/network layer to make each pool usable for the workloads it is good enough to run.
Physical infrastructure and power
Wuhu Yangtze River Delta compute centreChina Energy Engineering says the first phase included four 150MVA split-winding transformers and three 220kV outgoing lines. A 2026 procurement notice shows continued B-phase/7-5 construction activity.
Wuwei compute centreA second ByteDance/Volcano Engine campus in Wuwei has a planned 220kV substation with five 150MVA split-winding transformers and 140 10kV outgoing circuits. Environmental filings were active in 2026.
Thailand / Southeast AsiaThailand approved a $25bn TikTok data-infrastructure expansion in May 2026 across Bangkok, Samut Prakan and Chachoengsao. ByteDance's September financing also supports AI infrastructure including Southeast Asian data centres.
30 September research update — the VNET relationship is broader than Volcano Engine
WHAT CHANGED IN OUR UNDERSTANDING
The earlier page understated the ByteDance–VNET edge.
We previously treated the relationship mainly as a confirmed Volcano Engine × DYXnet cooperation. A 30 September Chinese-language procurement sweep found direct physical-infrastructure evidence explicitly naming ByteDance and VNET, while independent reporting around VNET’s largest 2026 wholesale award points to a much larger capacity relationship. The evidence now supports a broader relationship across network connectivity, physical fibre infrastructure and likely hyperscale data-centre demand — with an important boundary: VNET has still not publicly named ByteDance as the 510MW customer.
CONFIRMED PROCUREMENT EVIDENCE
ByteDance (Envision–VNET) fibre build · Ulanqab
June and July 2026 procurement notices explicitly name a “ByteDance (Envision–VNET) transmission pipeline and fibre-optic cable construction project”. The design notice specifies roughly 1.3 km of trenching, 48 DN100 galvanized-steel conduits and 32 × 288-core fibre cables totalling about 64 km; the later construction notice lists RMB18.378m of self-funded project capital. A 28 September subcontract notice shows physical works were still progressing.
STRONGLY REPORTED · NOT VNET-NAMED
~500MW ByteDance capacity report aligns closely with VNET’s 510MW disclosure
Bloomberg reporting, carried by Data Center Dynamics in March, said ByteDance had signed for about 500MW of VNET capacity in China. VNET subsequently disclosed two orders of 110MW and 400MW from one unnamed leading internet customer at separate Greater Beijing sites — 510MW in total, against only 517MW of total year-to-date orders at the time. The numerical and timing match is strong, but the customer identity remains unconfirmed by VNET.
CONFIRMED COMMERCIAL EDGE
Volcano Engine × DYXnet remains part of the picture
DYXnet, VNET’s wholly owned network-services subsidiary, separately disclosed cooperation with Volcano Engine across cloud computing, AI, security and network bandwidth. That relationship is now better understood as one layer of a wider ByteDance–VNET infrastructure overlap, rather than the only evidenced connection.
Research interpretation. We should no longer describe VNET exposure to ByteDance as merely a Volcano Engine ecosystem edge. The physical fibre project proves direct infrastructure coordination between named ByteDance and VNET endpoints, while the reported ~500MW award is consistent with VNET’s own unnamed 510MW hyperscale order. What we still cannot say is that VNET has formally confirmed ByteDance as that customer, or how much of ByteDance’s total domestic AI estate sits inside VNET facilities.
Economics: the ByteDance flywheel
Scroll horizontally to explore the full diagram →
China AI Infrastructure · chinaaievolution.com · See accompanying evidence and review date.
The flywheel is unusually powerful because internal demand can justify infrastructure engineering before external cloud customers arrive; Volcano Engine can then monetise that engineering externally. The key unresolved question is whether this improves total system cost per useful completed task after HBM premiums, additional networking, software complexity and power infrastructure are included.
Chinese market commentary names cooling, optics and power suppliers around ByteDance AI Rack projects, but much of it is not primary disclosure. We should not promote those names to confirmed graph edges until customer evidence is stronger.
How much ByteDance capacity sits with VNET?
The relationship is no longer just a research gap: direct ByteDance–VNET fibre construction is evidenced and a ~500MW capacity award is strongly reported. The remaining question is scale and attribution — which VNET campuses support ByteDance, how the 510MW is phased across sites, and what share of ByteDance’s wider domestic AI estate this represents.
Accelerator workload split
Which workloads actually run on Huawei vs Cambricon vs Iluvatar vs Nvidia, and at what utilisation/TCO? Supplier ranking is known; workload allocation is not.
Compute ↔ electricity scheduling
Inner Mongolia discussions are highly relevant, but they do not yet demonstrate hyperscale production workloads dynamically following power availability or price.
Method: primary company/government/infrastructure sources are preferred. Reuters sourcing is treated as corroborated reporting, not company confirmation. Vendor performance claims remain vendor-reported. Architectural inference is labelled separately from established commercial relationships.