Key Takeaways:
MiniMax's co-founder is betting the path to artificial general intelligence runs through model intelligence, not scale.
MiniMax co-founder Yeyi Yun said the company is increasingly confident of reaching $1 billion in annual recurring revenue by year-end, prioritizing model intelligence over proliferation as it defines AGI as the ability to generate 1% of global GDP.
"Enhancing the intelligence, rather than proliferation, of large AI models is our top priority," Yun said. "If AI can autonomously generate 1% of global GDP, it would likely have reached the AGI stage."
The guidance came as MiniMax-W (00100.HK) fell 3.5% on Sept. 1, with HK$621.34 million in short selling at a 6.668% ratio. The Shanghai-based lab reported a 283.1% year-over-year revenue surge to US$116.6 million in the first half of 2026, with gross profit improving 464.8% on lower infrastructure costs.
The $1 billion ARR target would mark a roughly ninefold increase from the current run rate, a bar that shows the stakes of the company's pivot toward intelligence-focused models such as the 428-billion-parameter M3.
MiniMax's strategic shift distinguishes it from rivals racing to ship ever-larger model families. The company's flagship M3, a 428-billion-parameter mixture-of-experts model with about 23 billion active parameters per token and a 1-million-token context window, prices at $0.30 per million input tokens and $1.20 per million output tokens — a fraction of OpenAI's GPT-5.3 Codex at $1.25 and $10.00, and Anthropic's Claude Opus 4.6 at $5.00 and $25.00. The M3 scores 80.5% on SWE-bench Verified, ahead of GPT-5.3 Codex's roughly 75%.
The pricing advantage is structural, not promotional. MiniMax's sparse-attention architecture cuts per-token compute by 28.4 times at a 1-million-token context, letting it undercut Western incumbents while protecting margins. That efficiency drove the 464.8% gross-profit improvement in the first half.
Yun's definition of AGI — autonomous generation of 1% of global GDP — frames the company's consumer-first strategy. Its Talkie companion app has drawn about 11 million monthly active users and 17 million downloads, feeding the multimodal interaction data that trains its models. The company serves more than 300 million users and over one million enterprise developers across 200 countries.
MiniMax-W listed on the Hong Kong Stock Exchange in January 2026, with shares surging 109% on the opening day to a market value above HK$100 billion (about $12.8 billion). The IPO raised HK$4.8 billion, with 90% of proceeds earmarked for research and development. Reaching the $1 billion ARR target would validate the premium valuation the market has assigned to the stock, which trades against a backdrop of heavy short interest. The company's open-weight distribution of its M-series models, alongside its Anthropic-compatible API, has made it a low-cost default for developers migrating from Claude, a channel that could accelerate the ARR ramp into 2027.
This article is for informational purposes only and does not constitute investment advice.
