At the Broadband Development Congress in Vienna during Network X 2026, Huawei Optical Business Product Line President Bob Chen outlined strategies to adapt network infrastructure for the artificial intelligence era. Chen stated that large AI models now feature approximately 10 trillion parameters, with token consumption increasing about 50-fold over the past year.
Three Pillars of Network Value
According to Huawei, adapting network value to the artificial intelligence era requires changes centered around three concept principles known as the three 'R's: rebuilding network architecture, reshaping business models, and redefining network operation and maintenance. Chen emphasized that network planning must align with AI service demands and AI data center (AIDC) layouts.
Huawei noted that network architecture must transition from administrative region-centric designs to AIDC-centric hierarchies comprising core, regional, and edge data centers. Highlighting the shift in traffic drivers, the company noted that cities like Zhongwei in China have become major national computing centers where smart computing traffic accounts for over 10 percent of the country's total bandwidth.
Operational Efficiency and Business Evolution
For enterprise segments, Huawei suggested evolving from single-dimension bandwidth sales to multi-dimensional services offering bandwidth, latency, and reliability supported by technologies like OXC, ASON, and quantum encryption. For home broadband, the company pointed toward full-house Wi-Fi solutions and intelligent smart-home applications powered by AI agents.
Huawei stated that artificial intelligence can significantly improve operational efficiency. The company cited AI-adjusted Wi-Fi that automatically detects interference sources to boost speeds by about 20 percent in noisy environments, alongside intelligent traffic analysis that reduces total device energy consumption by over 20 percent through dynamic power adjustments.
Key facts and where they come from
- Bob Chen stated that large AI models have approximately 10 trillion parameters.
Büyük yapay zeka modelleri artık yaklaşık 10 trilyon parametreye sahip
- Token consumption increased by about 50 times over the last year.
son bir yıl içinde token tüketimi yaklaşık 50 kat arttı
Read the original from Company press releases (PR Newswire) →