Goldman Sachs is positive on MiniMax's accelerating ARR and catalysts from the M3 and Hailuo 3 model launches

Overview

This report is Goldman Sachs' conference summary on MiniMax Group after the Asia Communacopia + Technology conference. The core content centers on ARR growth, changes in the API revenue mix, the M3 and Hailuo 3 model roadmap, unit economics, pricing room, and compute infrastructure advantages. The report maintains a positive view, believing MiniMax is one of the more competitive names among Chinese AI model companies.

Core views

Goldman Sachs believes MiniMax is well positioned to capture the expansion of the global AI model TAM, covering text/coding, multimodal, and agent/digital labor directions. The investment case mainly includes: rapid ARR ramp-up and rising API revenue mix; the upcoming M3 and Hailuo 3 launches may enhance product capability; the company has high API gross margins and potential room for price increases; and a proprietary compute cluster plus domestic chip adaptation may help lower costs and improve operating flexibility.

Analysis framework

The report uses a conference-notes analysis framework, based on management discussions at Goldman Sachs' conference, organizing by theme revenue trends, product iteration, unit economics, compute infrastructure, and risk factors, and combining Goldman Sachs' coverage framework to assign a Buy rating and target price.

Methodology notes

• Company researchConference notes

  Fundamental update driven by management discussions

  The report mainly comes from a summary of key points after MiniMax attended Goldman Sachs' Asia Communacopia + Technology conference, focusing on management's remarks on ARR, product roadmap, gross margin, pricing, and compute infrastructure.

• Valuation and ratingGoldman Sachs Buy/Neutral/Sell framework

  Total return potential relative to the coverage universe

  Goldman Sachs links Buy, Neutral, and Sell ratings to a stock's total return potential relative to the coverage universe; in this report MiniMax is rated Buy, with a 12-month target price of HK$1,000.

• Factor analysisGS Factor Profile

  Growth, financial returns, valuation multiples, and composite factors

  The appendix explains that Goldman Sachs' factor framework compares stocks with the market and peers through growth, financial returns, valuation multiples, and composite indicators, but the main text does not disclose MiniMax's specific factor percentile.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• MiniMax Group (0100.HK)

  Core coverage name

  Strengths

  Fast ARR growth and a rising API revenue mix; a clear roadmap for multimodal, coding, and agent models; API gross margin above peers; proprietary compute cluster and domestic chip adaptation provide cost and operating advantages.

  Weaknesses

  Profitability visibility still needs validation, and the business still depends on improved model performance and commercialization conversion.

  Comparison

  Goldman Sachs believes it is well positioned among Chinese AI model companies; Goldman Sachs estimates text API gross margin at about 40%, above the industry peer range of 20%-25%.

  Risks

  Model performance below expectations, commercialization below expectations, slower-than-expected profitability path, IP/content-generation risks, cash burn and self-funding capability, geopolitical risks amid U.S.-China technology competition.

• Open platform/API business

  Main ARR growth driver

  Strengths

  Revenue contribution has already exceeded half, gross margins are high, and it may benefit from pricing room after the intelligence level of the 3-series models improves.

  Weaknesses

  Price increases depend on model capability and customer willingness to pay.

  Comparison

  Goldman Sachs estimates text gross margin at about 40%, versus about 20%-25% for industry peers; multimodal gross margin is about 60%-70%.

  Risks

  Pricing pressure from competition and customer conversion and retention below expectations.

• Hailuo 3

  Product catalyst

  Strengths

  Expected to launch in June; uses an LLM-based native understanding-generation video model, and management sees strong commercial potential.

  Weaknesses

  Not yet launched, and actual performance and user adoption still need to be verified.

  Comparison

  The report notes its architecture can be compared with Seedance 2.0.

  Risks

  Launch delays, model performance below expectations, content-generation compliance, and IP risks.

Key data

• Report date2026-05-19The report header shows 19 May 2026 10:15AM HKT.
• April ARR growth100% increase versus FebruaryDriven mainly by API and the token plan.
• Open platform/API revenue contributionAbove 50%Around 30% at the end of last year, around 40% early this year, and now above half; the rest comes from Hailuo AI, MiniMax Agent, and Talkie/Xingye.
• M3 model launch timingSmaller version in the coming weeks, larger-parameter version in the coming monthsExpected to strengthen coding and agent capabilities.
• Hailuo 3 launch timingJune 2026To be launched after Hailuo 2.3 last October, positioned as an LLM-based native understanding-generation model.
• API text business gross marginGoldman Sachs estimates about 40%The report says industry peers are around 20%-25%.
• API multimodal business gross marginAbout 60%-70%Benefiting from cost efficiency and a flexible compute architecture.
• Domestic inference capabilityExpected to be fully available in 3Q26The company has already completed adaptation for major domestic chips.
• Target priceHK$1,000Goldman Sachs 12-month target price.
• RatingBuyGoldman Sachs maintains a Buy rating on MiniMax.

Impact & implications

If ARR growth, the M3 model, and the Hailuo 3 launch proceed as planned, MiniMax may further enhance API and multimodal commercialization capabilities and improve margins through higher pricing and cost efficiency. Proprietary compute and domestic chip adaptation may also reduce dependence on external compute supply and strengthen cost control. However, model performance, commercialization, the profitability path, and geopolitical tech competition remain key uncertainties.

Risks

• Model performance below expectations in global foundation-model competition.
• Slower-than-expected profitability visibility.
• Commercialization capability below expectations.
• IP and content-generation-related risks.
• Cash burn and self-funding capability risks.
• Geopolitical risks amid escalating U.S.-China technology competition.

What to watch

• Performance of coding and agent capabilities after the launch of the smaller-version M3 model in the coming weeks.
• Launch timing and performance improvement of the larger-parameter M3 version in the coming months.
• User feedback, video-generation quality, and commercialization progress after Hailuo 3 is released in June 2026.
• Whether API revenue mix can continue to rise and whether the token plan keeps driving ARR.
• Whether the improved intelligence level of the 3-series models leads to actual price increases and gross margin improvement.
• Progress in increasing the share of proprietary compute infrastructure by year-end.
• Whether domestic inference capability becomes fully available in 3Q26 after adaptation to major domestic chips.