Newshot

Tencent: More companies turning to domestic chips will speed up China's supply chain buildout

AI summary

Tencent reported Q3 revenue of RMB 207.336 billion, up 8% year on year, gross profit of RMB 115.911 billion, up 7%, and net profit attributable to shareholders of RMB 63.612 billion, up 0.8%. Management said capital expenditure is mainly going to AI infrastructure, which is not a one-off expense and can continue to support model training and token generation; operating pressure from this spending is expected to ease gradually by 2028, when procurement prices for similar equipment are expected to fall markedly and training and inference efficiency continues to improve. Compute supply remains a real constraint: tighter restrictions reduce available compute in the near term, affecting both leased and self-built capacity and putting pressure on model training, and the next-generation model may take longer to launch, while in the long run the restrictions may push more Chinese companies toward domestic supply chains and domestic chips, accelerating buildout of the domestic supply chain.

Translation(model translation, not the source's own text)

Tencent: More enterprises shifting to domestic chips will accelerate construction of the domestic supply chain

IT之家 November 19 news: Tencent released its third-quarter financial report yesterday. The report shows that in the third quarter it achieved revenue of 2073.36 亿元, up 8% year on year; gross profit of 1159.11 亿元, up 7% year on year; and net profit attributable to shareholders of 636.12 亿元, up 0.8% year on year.

According to 《科创板日报》, regarding continuously increasing AI investment, Tencent management said at the earnings call that the company's current capital expenditure is mainly used for AI infrastructure construction. This kind of investment is not a one-time expense; once built, the infrastructure can continue to be used for model training and Token generation. As infrastructure built in 2026 and 2027 takes effect in subsequent years, operating pressure brought by related investment is expected to gradually ease by 2028.

Management also mentioned that the investment cost of AI infrastructure is expected to show positive changes in the future. It believes that current equipment procurement prices are relatively high, and it is expected that by 2028, procurement prices for the same type of equipment will drop significantly; in addition, more importantly, the computational efficiency of model training and inference continues to improve, and under the same infrastructure conditions, it is expected to complete more training tasks and process more Tokens in the future.

Compute supply remains a real constraint facing the current AI business. Discussing the impact of the latest tightening restrictions, a Tencent executive said that in the short term, actually available compute will decrease, and both leased and self-built compute will be affected, while model training will also come under pressure. The precision adopted by Tencent's current Hunyuan Hy4 has relatively high training efficiency, but the launch of the next-generation model may take longer.

The executive also said that in the long term, related restrictions may push more Chinese companies to shift to the domestic supply chain and domestic chips, driving acceleration in the construction of the domestic supply chain, and related R&D resources will further concentrate in this field.

Separately, according to an IT之家 report last night, Tencent Holdings disclosed in its 2026 third-quarter earnings communication that, affected by chip restrictions, training the next-generation model will take longer. Tencent said that by increasing investment, exploring the upper limit of large-model intelligence upward, and building downward an AI application system covering personal life, office collaboration, and enterprise production, Tencent is accelerating its AI-driven transformation and moving toward becoming an AI-led technology company.

  • “Tencent: Affected by chip restrictions, training the next-generation model will take longer”

Related articles

  • Tencent: WorkBuddy's revenue in the last seven days exceeded $500 million annualized, and third-quarter revenue was more than 5 times that of the second quarter
  • Tencent executive: AI input-output efficiency is expected to improve by 2028, thanks to infrastructure reuse, hardware price cuts, and improved compute efficiency
  • Tencent: Affected by chip restrictions, training the next-generation model will take longer
  • Tencent刘炽平: After 微信 “小微” is fully launched, quarterly costs will be lower than 元宝 earlier this year
  • Tencent announces 2026 Q3 financial report: combined monthly active accounts of 微信 and WeChat 14.49 亿, up 2% year on year
  • Tencent WorkBuddy's daily active users exceed 1000 万, making it the AI product with the highest average daily Token usage among Chinese large language models

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by IT之家
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗