ByteDance Explores AI Chip Production With Samsung

• ByteDance is developing its own artificial intelligence chip and is in discussions with Samsung to manufacture it, according to multiple sources familiar with the project.
• The chip, designed for AI inference workloads, is part of the company’s broader push to secure advanced processors amid global supply constraints and U.S. export controls.
• The effort reflects a growing trend among major tech firms to build in‑house silicon to reduce reliance on external suppliers.

ByteDance Moves Toward In‑House AI Hardware

ByteDance is working on an AI chip and has entered negotiations with Samsung Electronics to manufacture it, according to people familiar with the matter. The company aims to receive sample units by the end of March and plans to produce at least 100,000 chips this year. One source said production could eventually scale to as many as 350,000 units. Discussions reportedly include access to memory chips, which remain in short supply due to the global surge in AI infrastructure demand.

A ByteDance spokesperson disputed the accuracy of information about the in‑house chip project but did not provide further details. Samsung declined to comment on the talks. If the collaboration proceeds, it would mark a significant milestone for ByteDance, which has been exploring chip development since at least 2022. The company began hiring semiconductor specialists that year as part of its long‑term strategy to support AI workloads.

ByteDance’s interest in custom silicon mirrors efforts by other global tech giants. Companies such as Google, Amazon and Microsoft have already developed their own AI chips to reduce dependence on Nvidia. These in‑house designs help optimize performance for specific workloads and provide more control over supply chains. ByteDance appears to be following a similar path as it expands its AI capabilities.

The project is internally codenamed SeedChip and forms part of a broader initiative to strengthen the company’s AI infrastructure. ByteDance sees custom hardware as essential for supporting its growing portfolio of AI‑driven services. These range from short‑video platforms to e‑commerce and enterprise cloud offerings. The company believes that AI will play a central role in shaping its future business.

Competition and Export Controls Drive Chip Development

Chinese technology companies face increasing pressure to develop their own AI chips due to U.S. export restrictions. These controls limit access to advanced processors needed for training and running large AI models. ByteDance’s rivals Alibaba and Baidu have already made significant progress in this area. Alibaba recently unveiled its Zhenwu chip for large‑scale AI workloads, while Baidu sells its Kunlunxin chips to external customers and plans to list the unit.

ByteDance has not yet released a commercial chip, but its efforts are accelerating. Reuters previously reported that the company was working with U.S. chip designer Broadcom on an advanced AI processor, with manufacturing planned at Taiwan’s TSMC. The new talks with Samsung suggest ByteDance is exploring multiple pathways to secure production capacity. Memory access is a key factor, as high‑performance AI chips require large amounts of fast memory.

The company’s AI ambitions extend beyond hardware. ByteDance founded a division called Seed in 2023 to develop large language models and promote their applications. These models support products such as the Doubao chatbot and its overseas version, Dola. Executives have acknowledged that ByteDance’s AI models lag behind global leaders like OpenAI, but they have pledged continued investment.

Zhao Qi, who oversees the company’s chatbot initiatives, told employees in January that AI investment would benefit all divisions. He emphasized that the company must continue to strengthen its AI capabilities to remain competitive. ByteDance sees AI as a transformative technology that will shape its long‑term strategy across multiple business units.

Massive AI Investment Signals Long‑Term Commitment

ByteDance intends to spend more than 160 billion yuan (about $22 billion) on AI‑related procurement this year, according to one source. More than half of that budget is expected to go toward purchasing Nvidia chips, including the H200 model. The remainder will support the development of the company’s in‑house chip. This level of investment underscores the scale of ByteDance’s AI ambitions.

The company’s reliance on Nvidia highlights the challenges of building a fully independent hardware ecosystem. Even with custom chip development underway, ByteDance still depends heavily on established suppliers. The SeedChip project aims to reduce that dependence over time. Scaling production to hundreds of thousands of units would help the company secure more predictable access to critical components.

ByteDance’s push into AI hardware reflects a broader industry trend. As AI workloads grow more complex, companies are seeking specialized chips optimized for inference and training tasks. Custom processors can offer performance advantages and cost savings. They also provide strategic benefits in an environment shaped by geopolitical tensions and supply chain constraints.

The company’s long‑term strategy appears to combine in‑house development with partnerships across the global semiconductor industry. Whether the Samsung negotiations lead to a formal manufacturing agreement remains to be seen. For now, the talks highlight ByteDance’s determination to secure the hardware it needs to compete in the rapidly evolving AI landscape.

Samsung has become an increasingly important partner for companies seeking alternatives to TSMC for advanced chip manufacturing. Its foundry division has invested heavily in next‑generation process technologies, including 3‑nanometer production. This makes Samsung an attractive option for firms looking to diversify their supply chains amid global semiconductor uncertainty.