
MediaTek believes AI is moving to teams of agents.
By 2030, 1 million H100s will be mid-tier.
This signals massive compute growth.
What happened
At DIGITIMES' AI on Chips forum, MediaTek senior director Bor-Sung Liang said AI development is shifting from individual large language models (LLMs) toward collaborative teams of AI agents with different roles.
Why it matters
Liang predicted that by 2030, a computing cluster with 1 million H100 GPUs will be considered mid-tier, reflecting the massive scale of compute expected in the AI race.
What to watch
The shift to multi-agent AI systems implies greater demand for specialized chips and infrastructure, potentially reshaping semiconductor industry priorities.
Ask the AI about this article →
The article reports on comments by MediaTek's Bor-Sung Liang at a DIGITIMES forum. He describes a transition in AI from single LLMs to systems of cooperating AI agents. This shift, if realized, would require vastly more computing power, as implied by his prediction that 1 million H100s would be considered mid-tier by 2030. Such a scale-up suggests significant implications for chip makers and data center operators, though the article does not provide further details on market impact or specific strategies.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Nvidia is reportedly in discussions to invest in AI startup Perplexity as part of a funding round that would v…

SK Hynix is pushing beyond HBM (high-bandwidth memory, a key component in AI chips) to two new areas: HBF and…

dotData explained its approach to using a separate LLM (an AI that understands and generates text) to rate ano…

ITmedia's ICT Research Division tested ChatGPT, Gemini, and Claude Sonnet 5 in July 2026

Anthropic's Claude AI service is currently experiencing an outage, as indicated by the article title

Taiwan prosecutors indicted nine people, including employees of Nvidia and Super Micro, for allegedly exportin…
