🧔♂️ A friendly human may check it before it goes live. More news here
South Korea’s Upstage said to hire securities firms for 2026 IPO
Upstage, a South Korean AI startup, has appointed KB Securities and Mirae Asset Securities to prepare for a potential IPO as early as the second half of 2026, according to people familiar with the matter.
The Seoul-based firm may list on the Korean exchange by the first half of 2027.
Upstage is among five teams shortlisted by the South Korean government to develop a national foundational AI model, with the group to be narrowed down to two finalists.
The company, co-founded in 2020 by Sung Kim, offers enterprise document processing and large-language model products.
The firm has raised about US$157 million from backers including Korea Development Bank, Amazon, AMD, and SBVA (formerly SoftBank Ventures Asia).
Upstage would be the first generative AI startup in South Korea to go public since ChatGPT’s launch.
🔗 Source: Bloomberg
🧠 Food for thought
Implications, context, and why it matters.
South Korea’s national AI project limits near-term IPO prospects for Upstage
- Staged government reviews will narrow the field to a few finalists 1. The timetable overlaps with IPO prep, so Upstage faces elimination risk before a listing that underwriters will factor into valuation.
- Upstage’s financials remain undisclosed. No public revenue or Annual Recurring Revenue (ARR) data. No profitability figures. No enterprise customer counts beyond “Fortune 500 companies including Samsung” 2. Pricing a 2026 to 2027 IPO stays speculative, given reliance on government GPU resources that could slow development 1.
- Appointing underwriters starts preparation. Korean IPOs need audited financials and investor roadshows (marketing meetings with potential investors). H2 2026 looks tight even if Upstage advances through the government process and shows commercial traction.
Infrastructure vendors can target Korea’s national AI program participants before finalist selection
- The government is offering 200 billion won (about $136 million) in support to multiple consortia (multi-organization teams) in the national foundational model effort 3. The package covers high performance GPUs, data and labor costs. Vendors with compute or curated datasets can pitch now. They can also offer model testing tools or integration services while budgets are open and architectures are still forming.
- Each team brings multiple academic and industry partners 3, which creates several entry points for deals. Reports cite instability with some government GPU resources 1. That gives alternative compute suppliers a shot to win on reliability and support.
Recent Upstage developments
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.




