10 Jun 2026
CMB International Global Markets | Equity Research | Initiation
Minimax Group (100 HK)
Evolving into a one-stop AI platform
Minimax is one of the leading global AI foundational model companies. The company
differentiates itself from other model companies in terms of: 1) its commitment to
developing multi-modal models from day one; 2) its cost-efficient approach to deliver
AI intelligence; 3) its capabilities to reach and monetize on global user base.
Minimax’s total revenue reached US$ in 2025, up by 159% YoY; and AI
Native Products/Open Platform and Other AI-based Enterprise Services each account
for 67%/33% of total revenue. Minimax currently has a relatively small market share
of % in terms of global model- based revenue, according to CIC. That said,
supported by its SOTA multi-modal model capabilities and diversified AI applications,
we expect Minimax to grow at a 181% revenue CAGR over FY25-28E and further gain
market share in the global foundation model market. We value Minimax at HK$
per share based on 80x FY26E PS. Initiate at BUY.
◼ Strong competitive moats across multiple aspects. 1) Multi-modal capabilities:
Minimax’s models ranked the top 15 across text, video, and speech benchmarks
(per Artificial Analysis). And we expect Minimax to maintain its competitiveness
across multiple modalities, underpinned by its strong R&D capabilities, talent
density and efficient organizational structure;
2) strong cost efficiency: MiniMax M Series strike an optimal balance between
performance and costs. Minimax M3 and are in the most attractive quadrant
of cost to run intelligence as per Artificial Analysis; 3) capabilities to reach global
user base: MiniMax has launched all its foundational models and products across
international markets. Revenue from Mainland China/overseas market accounted
for 27%/73% of total revenue in FY25.
◼ Enormous addressable market for foundational model providers. The global
foundation model market remains at early development stage with strong growth
potential. Per CIC, the global foundation model market is expected to grow at a
CAGR of 65% over 2025-2029E to US$ in 2029E, with application/MaaS
market accounting for 73%/27% respectively, primarily driven by the
technological advancement and cost reduction of foundation model services.
Minimax currently has a relatively small market share of % in terms of global
model-based revenue, according to CIC. That said, supported by its SOTA multi-
modal model capabilities and diversified AI applications, we expect Minimax will
continue to gain market share in the global foundation model market.
◼ Total revenue to grow at a 181% CAGR over FY25-28E. Driven by he strong
demand for AI Native Products and Minimax’s multi-modal capabilities, we
expect Minimax’s total revenue to grow at a 2025-28E CAGR of 181% and reach
US$1,748mn in FY28E. Supported by the operating leverage and enhanced
efficiency, we expect adjusted net margin to improve from -317% in FY25 to -19%
in FY28E. We value Minimax at HK$ per share (US$) based on 80x
FY26E PS. Our target PS multiple is at a premium to the global peers, mainly
reflecting: 1) Minimax’s robust revenue growth outlook underpinned by the
booming AI demand; 2) Minimax’s strong multi-modal capabilities and inference
cost efficiency.
Earnings Summary
Target Price HK$
Up/Downside %
Current Price HK$
China Software & IT Services
Saiyi HE, CFA
(852) 3916 1739
hesaiyi@
Wentao LU, CFA
luwentao@
Ye TAO, CFA
(852) 3850 5226
franktao@
Shuyin GUO
(852) 3916 3716
guoshuyin@
Stock Data
Mkt Cap (HK$ mn) 145,
Avg 3 mths t/o (HK$ mn) 1,
52w High/Low (HK$) NA/NA
Total Issued Shares (mn)
Source: FactSet
Shareholding Structure
Yan Junjie %
Alibaba %
Source: HKEx
Share Performance
Absolute Relative
1-mth % %
3-mth % %
6-mth NM NM
Source: FactSet
12-mth Price Performance
Source: FactSet
Revenue (US$ mn) 31 79 285 743 1,748
Adjusted net profit (US$ mn) () () () () ()
EPS (Adjusted) (US$ cents) () () () () ()
P/S (x)
Source: Company data, Bloomberg, CMBIGM estimates
PLEASE READ THE ANALYST CERTIFICATION AND IMPORTANT DISCLOSURES ON LAST PAGE 1
(YE 31 Dec) FY24A FY25A FY26E FY27E FY28E
Contents
Investment thesis .........................................................................................3
One of the leading multi-modal foundational model providers .....................................................3
Strong competitive moats across multiple aspects.........................................................................3
Enormous addressable market for foundational model providers..................................................3
Total revenue to grow at a 181% CAGR over FY25-28E .............................................................3
Minimax: leading multi-modal foundational model providers...............4
Multi-modality: offer SOTA models across multiple modalities ..................................................5
Cost optimization: strong cost efficiency of model training and inference ...................................7
Overseas expansion: monetize on global user base .......................................................................8
Business model: wide range of AI-native products built on SOTA
foundational models..................................................................................10
Minimax: intelligent agent application........................................................................................10
Hailuo AI: Visual Generation Platform.......................................................................................12
MiniMax Audio: Audio Generation Tool....................................................................................13
Talkie/Xingye: AI-powered Multi-modal Entertainment Platform .............................................14
Open Platform..............................................................................................................................14
Industry analysis: enormous addressable market for foundational model
providers....................................................................................................16
Competitive landscape.................................................................................................................18
Minimax: financial forecast and valuation.............................................20
Operating efficiency and margin analysis ...................................................................................21
Valuation .....................................................................................................................................22
Appendix....................................................................................................23
Investment risks ........................................................................................24
Fierce competition in the LLM industry......................................................................................24
Relatively low floating share base ...............................................................................................24
Continuous investment in R&D may affect profitability and cash flow .....................................24
Failure to attract and retain key talents may affect R&D and operation......................................24
Evolving regulations may adversely affect business operation ...................................................24
Investment thesis
One of the leading multi-modal foundational model providers
Minimax is one of the leading global AI foundational model companies. The company
differentiates itself from other model companies in terms of: 1) its commitment to developing
multi-modal models from day one; 2) its cost-efficient approach to deliver AI intelligence; 3) its
capabilities to reach and monetize on global user base. Minimax’s total revenue reached
US$ in 2025, up by 159% YoY; and AI Native Products/Open Platform and Other AI-
based Enterprise Services each account for 67%/33% of total revenue. Minimax currently has a
relatively small market share of % in terms of global model-based revenue, according to CIC.
That said, supported by its SOTA multi-modal model capabilities and diversified AI applications,
we expect Minimax will continue to gain market share in the global foundation model market.
Strong competitive moats across multiple aspects
Minimax has established three major competitive edges: 1) multi-modal capabilities: Minimax’s
models excel across various modalities. According to Artificial Analysis, Minimax’s models
ranked the top 15 across text, video, and speech benchmarks. And we expect Minimax to maintain
its competitiveness across multiple modalities, underpinned by its strong R&D capabilities, talent
density and efficient organizational structure; 2) strong cost efficiency of its models: MiniMax M
Series strike an optimal balance between performance and costs, by adopting an advanced model
architecture. The API pricing of MiniMax-M2 was US$ per million input tokens and US$
per million output tokens, equivalent to only % of the price of leading overseas models.
Minimax and M3 further enhanced its performance and cost efficiency: according to
Artificial Analysis, MiniMax-M3 and are in the most attractive quadrant of cost to run
intelligence; 3) capabilities to reach global user base: MiniMax has launched all its foundational
models and products across international markets. Revenue from Mainland China/overseas market
grew by 132%/171% YoY to US$ in 2025, each accounting for 27%/73% of total
revenue.
Enormous addressable market for foundational model providers
The global foundation model market remains at early development stage with strong growth
potential. Its total market size reached US$ in 2025, with application/MaaS market
accounting for 66%/34% respectively, according to CIC. Looking ahead, the global foundation
model market is expected to grow at a CAGR of 65% over 2025-2029E, to US$ in 2029E,
with application/MaaS market accounting for 73%/27% respectively, primarily driven by the
technological advancement and cost reduction of foundation model services. Minimax currently
has a relatively small market share of % in terms of global model-based revenue, according to
CIC. That said, supported by its SOTA multi-modal model capabilities and diversified AI
applications, we expect Minimax will continue to gain market share in the global foundation
model market.
Total revenue to grow at a 181% CAGR over FY25-28E
Driven by the strong demand for AI Native Products and Minimax’s multi-modal capabilities, we
expect Minimax’s total revenue to grow by 260%/161%/135% YoY to US$285/743/1,748mn in
FY26/27/28E. AI Native Products/Open Platform and Other AI- based Enterprise Services will
each account for 45%/55% of total revenue in FY28E, vs. 67%/33% in FY25. Supported by the
operating leverage and enhanced efficiency, we expect adjusted net margin to improve from -
317% in FY25 to -148%/-64%/-19% in FY26/27/28E. We value Minimax at HK$ per share
(US$) based on 80x FY26E PS. Our target PS multiple is at a premium to the global peers,
mainly reflecting: 1) Minimax’s robust revenue growth outlook (25-28E CAGR of 181%)
underpinned by the booming AI demand; 2) Minimax’s strong multi-modal capabilities and
inference cost efficiency.
Minimax: leading multi-modal foundational model providers
Founded in 2022, Minimax is one of the leading global AI foundational model companies. The
company differentiates itself from other model companies in terms of: 1) its commitment to
developing multi-modal models from day one; 2) its cost-efficient approach to deliver AI
intelligence; 3) its capabilities to reach and monetize on global user base.
The company has built three product layers: 1) the foundation model layer offers intelligence
support to Minimax’s AI native products and Open Platform, including large language models
(MiniMax M Series), video generation models (Hailuo), and models for speech and music
generation (Speech-02); 2) the AI-native product layer offers AI experience to individual users,
developers and enterprises, including intelligent agent application (MiniMax), visual generation
platform (Hailuo AI), audio generation tool (MiniMax Audio), and multi-modal entertainment
platform (Talkie/Xingye); 3) Minimax Open Platform: offers scalable and configurable AI services
to enterprise customers and developers through public APIs and services.
Figure 1: Minimax: product overview
MiniMax
Intelligent Agent Application
Hailuo AI
Visual Generation Platform
MiniMax is an intelligent AI agent application, which is
designed to autonomously perform a wide range of tasks through
natural language instructions. Supported by Minimax’s foundation
models, MiniMax Agent can plan, reason, and execute complex
actions such as coding, research, document drafting, and
presentation creation within a unified workspace.
Hailuo AI fully integrates Hailuo-02 model. It is offered in both
web and app forms, and is designed for real-time, high-quality
image and video generation.
1) In-app top-up
2) Subscriptions
1) In-app top-up
2) Subscriptions
MiniMax Audio
Audio Generation Tool
MiniMax Audio is designed to provide users with high-fidelity
audio generation capabilities. Accessible via web platform,
MiniMax Audio integrates the company’s Speech-02 model to
support interactive audio synthesis and generate natural, high-
quality speech from text input.
1) In-app top-up
2) Subscriptions
Talkie/Xingye
Multi-modal Entertainment
Platform
MiniMax Open Platform
Talkie (for international markets)/Xingye (for domestic market) is a
globally recognized AI-native multi-modal entertainment platform.
Users of Talkie/Xingye can engage with emotionally responsive AI
themes or virtual characters powered by the company’s proprietary
AI-models.
Open Platform offers scalable, configurable AI services to
enterprise customers and developers across more than 100 countries
and regions. Through public APIs and services, enterprise and
developer customers can access the company’s foundation models
and integrate such text, video and audio model capabilities into
their products and services.
1) In-app top-up
2) Subscriptions
3) Online marketing service
1) Token/Project-Based
2) Subscription
MiniMax M Series
Large Language Model
Hailuo-02
Video Generation Model
Speech-02
Speech Generation Model
The MiniMax M Series, comprising MiniMax-M1 and MiniMax-
M2, represents Minimax’s flagship family of large language models.
focused on improving performance on complex, real-
world tasks, particularly in coding and workplace NA
scenarios. achieved globally leading performance across
key productivity scenarios, including coding, tool use and workplace
applications.
The Hailuo-02 series model generates high-quality video content from
a variety form of information inputs. Hailuo-02 NA
offers cinematic video quality, advanced prompt adherence,
smooth motion, and style diversity.
The Speech-02 model series is designed to generate natural,
high-quality speech from text input. Speech-02 model delivers NA
hyper-realistic, personalized voice synthesis across multiple
languages.
AI-native product offerings
MonetizationIntroductionProducts
Foundation Model Suite
Music-02
Music Generation Model
MiniMax Music is the music synthesis model designed for music
composition, accompaniment, and background audio NA
generation.
Image-01
Video Generation Model
Image-01 model enables cinematic quality image generation
from text prompts. It supports character styling and other NA
creative compositions.
Source: Company data, CMBIGM
Minimax’s total revenue reached US$ in 2025, up by 159% YoY. By segment, revenue
from AI Native Products/Open Platform and Other AI-based Enterprise Services grew by
143%/198% to US$ accounting for 67%/33% of total revenue. By specific
product, revenue from Talkie & Xingye/Hailuo AI/Open Platform and Other AI- based Enterprise
Services/Minimax Voice/Minimax each represented 35%/33%/29%/2%/1% of total revenue in
9M25.
Figure 2: Minimax: revenue trend Figure 3: Minimax: revenue mix (9M25)
(US$mn) MiniMax
90
80
70
60
50
40
30
20
10
0
2023 2024 2025
900%
800%
700%
600%
500%
400%
300%
200%
100%
0%
Open Platform and
Other AI-based
Enterprise Services
29%
1%
Hailuo AI
33%
Minimax
Open Platform and Other AI-based Enterprise Services AI
Native Products
YoY (RHL)
Talkie/Xingye
35%
Voice
2%
Source: Company data, CMBIGM Source: Company data, CMBIGM
Multi-modality: offer SOTA models across multiple modalities
Minimax is one of the first and a few foundational model developers that have first-party models
across all modalities, including language, speech, image, and video models. By integrating multi-
modal capabilities, Minimax is capable of offering AI-native products that can handle complex and
multi-dimensional interactions across text, visual and audio. The multi-modal models support
Minimax to achieve higher level of AI intelligence, as real-world human interaction is inherently
multi-modal.
Figure 4: Minimax: one of a few players with first-party models across all modalities.
Source: Artificial Analysis, CMBIGM
Moreover, Minimax’s models excel across various modalities. According to Artificial Analysis,
Minimax’s models ranked the top 15 across text, video, and speech benchmarks. And we expect
Minimax to maintain its competitiveness across multiple modalities, underpinned by its strong
R&D capabilities, talent density and efficient organizational structure.
Figure 5: Minimax: … and excels in multiple modality
Source: Artificial Analysis, CMBIGM Note:
data as of 8 Jun 2026
To fully utilize the multi-modal capabilities of Minimax’s models, Minimax introduced MMX- CLI
in Apr 2026, an all-modal command-line tool developed for AI agents. Agents can directly invoke
it in environments such as Claude Code or OpenClaw without building an MCP server, supporting
for text, image, video, speech, music generation, and visual understanding capabilities.
Figure 6: Minimax: MMX-CLI
Source: Company data, CMBIGM
Minimax M3, Minimax’s latest foundational model, is a natively multimodal model that has
undergone mixed-modality training from Step 0 and supports muti-modal input. This training
approach, with text and other modalities interleaved within a sequence, allows Minimax M3 to: 1)
achieve deep alignment between textual and visual semantic spaces; 2) scale the training data to the
order of 100 trillion tokens.
Cost optimization: strong cost efficiency of model training and inference
MiniMax M Series models also strike an optimal balance between performance and costs, by
adopting an advanced architecture. The API pricing of MiniMax-M2 was US$ per million
input tokens and US$ per million output tokens, equivalent to only % of the price of
leading overseas models. And running agents on Minimax continuously for one hour at an
output speed of 100 tokens per second costs only US$1, and 4 agents can operate continuously for
an entire year with a budget of US$10,000. Minimax and M3 further enhanced its
performance and cost efficiency, underpinned by the innovation in architecture: according to
Artificial Analysis, MiniMax-M3 and are in the most attractive quadrant of cost to run
intelligence.
Figure 7: Minimax-M3 & in the most attractive quadrant of cost to run intelligence
Source: Artificial Analysis, CMBIGM Note:
data as of 8 Jun 2026
To improve context scaling and agentic capabilities, Minimax M3 model innovated on the
attention mechanism and introduced Minimax Sparse Attention (MSA) architecture. The MSA
architecture can partition the KV into blocks more precisely, achieving higher effective context
coverage. This allows M3's per-token compute to be just 1/20 that of the previous- generation
model. Minimax M3 Model achieved a speed up of over 9x in the prefilling stage and over 15x in
the decoding stage, while the MSA also matched full attention on the vast majority of capabilities.
Figure 8: Minimax Sparse Attention
Source: Company data, CMBIGM
Overseas expansion: monetize on global user base
MiniMax has launched all its foundational models and products across international markets from
day one, which enables the company to capture commercial opportunities in an enormous global
foundation model market of over US$300bn. By geography, revenue from Mainland
China/overseas market grew by 132%/171% YoY to US$ in 2025, each accounting
for 27%/73% of Minimax’s total revenue.
Figure 9: Minimax: revenue trend (by geography)
(US$ mn) 60
Figure 10: Minimax: revenue mix (9M25, by
geography)
Mainland
50 Others
28%
40
China
27%
30
20
10
0
2023 2024 9M25 United Singapore
Others United States Singapore Mainland China
States
20%
24%
Source: Company data, CMBIGM Source: Company data, CMBIGM
Individual (B2C) Text, Video, Speech and Multi-modal Freemium, subscriptions, online marketing
Music Entertainment Platform service, in-app purchasesTalkie/Xingye
Business model: wide range of AI-native products built on SOTA
foundational models
Based on its multi-modal foundation model suite, Minimax offers a wide range of AI-native
applications and products, including Minimax Chat/Agent, HailuoAI, Minimax Audio,
Talkie/Xingye and Open Platform. Its products target both consumers and enterprises, monetizing
through consumer subscriptions, token-based in-app purchases, online marketing service and API
monetization.
Figure 11: Minimax: AI-native product offerings
MiniMax Individual (B2C) Text, Video, Speech and
Music
Intelligent Agent
Application
Freemium, subscriptions, token-based in- app
purchases
MiniMax Audio Individual (B2C) Speech and Music Audio Generation Tool Freemium, subscriptions, token-based in-
app purchases
Open Platform Enterprise and
Developer (B2B)
Text, Video, Speech and
Music
API Platform Freemium, token-based API billing,
subscription, enterprise licensing
Source: Company data, CMBIGM
Minimax: intelligent agent application
MiniMax is an intelligent AI agent application, which is designed to autonomously perform a wide
range of tasks through natural language instructions. Supported by Minimax’s foundation models,
MiniMax Agent can plan, reason, and execute complex actions such as coding, research, document
drafting, and presentation creation within a unified workspace.
Figure 12: Minimax: agent application
Source: Company data, CMBIGM
Product name Target users Key underlying models Primary use cases Monetization model
Freemium, subscriptions, token-based in- app
purchases
Flagship Visual
Generation Platform
Individual (B2C) VideoHailuoAI
MiniMax has reached the level of SOTA models in various areas: 1) software engineering:
On SWE-Pro, which covers multiple programming languages, achieved a % accuracy
rate, matching -Codex. It even delivers higher score on benchmarks closer to real-world
engineering scenarios, such as SWE Multilingual () and Multi SWE Bench (); 2) office
collaboration: In the GDPval-AA evaluation, which measures this capability to possess
professional knowledge across various fields, achieved an ELO score of 1,495 among 45
models, second only to Opus , Sonnet , and , and surpassing ; 3)
entertainment: Based on , Minimax built OpenRoom, an interaction system based on an
agent harness that liberates AI interaction from plain text streams and places it within a Web GUI
space where everything is interactive.
Figure 13: Minimax: agent use cases
Source: Company data, CMBIGM
The company upgraded Minimax Agent platform to Mavis in May 2026, introducing several new
functions: 1) Minimax Agent Teams, which allows multiple agents with skills to work in parallel,
significantly improving the agents’ capability to complete complicated tasks; 2) merging Token
Plan and Agent Plan, which grant subscribers access to all Minimax’s offerings including agents,
, video, music and audio models. The upgrade further enhanced Minimax agentic capabilities
and drove up token consumption.
Figure 14: Minimax Agent Teams
Source: Company data, CMBIGM
Monetization: MiniMax Agent employs a tiered monthly subscription model, featuring the
“Basic” plan at US$ per month and the “Pro” plan at US$ per month. These premium
subscriptions include exclusive benefits such as peak-hour priority access and early access to beta
features. Additionally, users may purchase task-execution credits at US$ per 5,000 credits,
which can be applied to execute supplementary tasks via the MiniMax Agent. In addition,
MiniMax Agent offers a team plan under which users are charged US$15 per seat per month,
providing each team with a shared monthly credit pool and enhanced features.
Hailuo AI: Visual Generation Platform
Hailuo AI is Minimax’s flagship visual generation platform, designed for real-time, high-
definition video and image synthesis. Hailuo AI serves creators, advertisers, and everyday users,
offering intuitive tools for crafting visual content through text and image inputs.
Monetization: Hailuo AI employs a tiered subscription model, with plans starting at US$ per
month for the “Standard” tier and scaling to US$ per month for the “Max” tier. Premium
subscriptions unlock more functionality including 1080p resolution outputs, watermark-free
outputs, and priority access to new features. Complementing subscription tiers, users may purchase
top-up credits within the Hailuo AI.
Figure 15: Minimax: Hailuo Video
Source: Company data, CMBIGM
MiniMax Audio: Audio Generation Tool
MiniMax Audio is Minimax’s audio generation tool, which provides users with high-fidelity
speech and music generation capabilities. Within the MiniMax Audio platform, users can also
generate speech using the Speech-02 model by typing custom text across more than one language.
They can select voices from a curated library of presets with different emotions and styles, adjust
pitch and speed, and manage their own voice profiles, and preview/export voice outputs.
Monetization: MiniMax Audio adopts a credit-based consumption model, priced at US$ per
million credits. Additionally, users may choose from tiered monthly subscription plans from
“Starter” plan priced at US$ per month to “Pro” plan priced at US$ per month. These
premium subscriptions offer benefits such as accelerated speech generation, the ability to generate
speech with specified emotions and languages, and access to an expanded library of more than 100
distinct voices.
Figure 16: Minimax: Minimax Audio
Source: Company data, CMBIGM
Talkie/Xingye: AI-powered Multi-modal Entertainment Platform
Talkie (for international markets) and Xingye (for Chinese domestic market) are emotionally
intelligent AI-native multi-modal entertainment platform designed for real-time human-AI
interaction experience. Talkie/Xingye enables users to co-create, customize, and interact with
virtual themes and characters that exhibit memory, emotion, and dynamic personality.
Monetization: Talkie/Xingye have various monetization methods including online marketing
services, subscription services and in-app purchases. Users may purchase in- app credits which can
be spent in-app items. Additionally, a monthly membership subscription offers enhanced features
such as more chat reply regenerations, enhanced chat memory and faster response. Additionally,
Minimax generates online marketing service revenue through Talkie, which offers integrated
marketing and promotional services.
Figure 17: Minimax: Talkie/Xingye
Source: Company data, CMBIGM
Open Platform
Minimax Open Platform offers scalable, customizable AI services to global enterprise customers.
Through public APIs, services and cloud SDKs, enterprise and developer customers can access
Minimax’s foundation model suite and integrate Minimax’s foundation model into their own
products and services without the need for independent foundation model development.
Monetization: Minimax Open Platform mainly monetized through usage-based pricing, where
fees are charged per “token”, “video clips”, or “characters”. Subscription tiers provide fixed-fee
access, ranging from entry-level plans for testing to premium options with enhanced features. The
company also launched Coding Package, offering tiered subscription plans designed for
programming and software-development scenarios. Minimax also provides enterprise customers
with other AI-based enterprise services, mainly consists of arrangements customized to enterprise
requirements and licensed deliverables.
Figure 18: Minimax: Open Platform
Source: Company data, CMBIGM
Industry analysis: enormous addressable market for
foundational model providers
Since the GPT-3 launched in 2020, the foundation model industry has undergone rapid
development, significantly improving the model intelligence and accelerating monetization of AI
applications:
1) expansion of model scale: OpenAI’s GPT-3 was launched in 2020 with only 175bn parameters,
while more advanced and latest models like Gemini 3 and significantly outperformed it
with much larger model scale and better intelligence. This improvement was primarily thanks to
the introduction of Mixture-of-Experts (MoE) architecture, which allows foundation models to
expand scale while keeping computational cost and latency low.
2) improving context windows and reasoning efficiency: Claude’s 100,000-token context window
enables interaction with ultra-long documents, compared to GPT-3 which only supported a context
window of ,048 tokens. Besides, the improvement to attention mechanism further improves the
reasoning efficiency, allowing models to process long context at reasonable costs.
3) alignment with humans: RLHF (reinforcement learning from human feedback) allows
foundation models to be more receptive to user prompts, enhances response quality and improve
accuracy.
4) emergence of CoT (chain-of-thought) and reasoning models: the CoT prompting technique and
reasoning models lead to a divergence in future model development: one class of models will be
optimized for fast and factually accurate responses, while the other will focus on deeper and more
resource-intensive reasoning.
As the intelligence level of foundation model continues to advance, AI currently achieves the level
of Agents, which can take actions based on users’ prompt.
Figure 19: Level of AI intelligence
Level Name Description
L1 Chatbots AI with conversational language
L2 Reasoners AI with human-level problem solving
L3 Agents AI that can take actions
L4 Innovators AI that can aid in invention
L5 Organizations AI that can do the work of an organization
Source: OpenAI, CMBIGM
Rapid advancement of foundation model accelerates applications and monetization of foundation
models. Major applications of foundation models currently include: 1) productivity, such as
general chatbots and coding tools, which covers a wide range of use cases of work and daily life like
information search, writing, coding, education and business analysis etc. Leading products include
ChatGPT, Grok, Claude Code and Cursor; 2) entertainment, which covers highly diverse use cases
spanning role-play, companionship, and everyday Q&A interactions etc. Leading products include
Character AI and Minimax’s Talkie/Xingye; 3) visual generation which primarily includes image
generation and video generation. Leading products include Google’s Nano Banana, Midjourney,
OpenAI’s Sora, Kuaishou’s Kling and Minimax’s Hailuo AI; 4) audio generation, which is a
universal interface of interaction in the AI era, connecting AI capabilities to a broad downstream
application market. Leading products include OpenAI, MiniMax, and ElevenLabs; 5) enterprise
services, which include the model capabilities, toolkits and industry solutions offered by the
foundational model companies.
Figure 20: Market landscape of foundation model applications
Source: CIC, CMBIGM
The global foundation model market remains at early development stage. Its total market size
reached US$ in 2025, with application/MaaS market accounting for 66%/34% respectively,
according to CIC. Looking ahead, the global foundation model market is expected to grow at a
CAGR of 65% over 2025-2029E, to US$ in 2029E, with application/MaaS market
accounting for 73%/27% respectively, primarily driven by technological advancement and cost
reduction of foundation model services.
Figure 21:Global: foundation model market size
(USDbn)
250
MaaS Application
200
150
100
50
0
2023 2024 2025E 2026E 2027E 2028E 2029E
Source: CIC, CMBIGM
Competitive landscape
Minimax currently has a relatively small market share of % in terms of global model- based
revenue, according to CIC. That said, supported by its SOTA multi-modal model capabilities and
diversified AI applications, we expect Minimax will continue to gain market share in the global
foundation model market.
Figure 22: Global foundation model: market share by model-based revenue (2024)
35%
30%
25%
20%
15%
10%
5%
0%
Source: CIC, CMBIGM
Minimax launched Minimax M3 model in Jun 2026, the first open-weight model with three SOTA
capabilities: 1) Coding and Agentic: Minimax M3 scored % on SWE Bench pro, surpassing
(%) and Gemini Pro (%), which demonstrated its strong capability in coding
and automated workflows; 2) 1M tokens context window: Powered by the proprietary MiniMax
Sparse Attention (MSA) architecture, Minimax M3 API supports up to 1M tokens context window
with a guaranteed minimum of 512K tokens. The 1M context is the infrastructure for long-range
Agent tasks, long-range Coding, and long- video understanding; 3) Native multimodality:
Minimax M3 is a natively multimodal model, with multimodal training from step zero and
pretraining data up to 100T+. The model supports image and video input, and can operate on a
desktop computer.
Figure 23: Minimax M3: benchmark performance
Source: Company data, CMBIGM
According to Artificial Analysis, Minimax-M3 ranked the top 10 among global frontier models by
intelligence index. It also ranked the second among models developed by Chinese vendors, leading
other models like Kimi , -Pro, DeepSeek V4 Pro and .
Figure 24: Frontier models: intelligence index comparison
Source: Artificial Analysis, CMBIGM
Supported by its innovative architecture and strong cost efficiency, Minimax M3 model continued
to maintain its pricing advantages vs. Qwen , equivalent to only % of
that of the . Meanwhile the blended API pricing of M3 model increased by 2-3 times
compared to model, which should support GPM expansion of the company in our view.
Figure 25: Global frontier models: API pricing comparison
Region Provider Model US$/1M tokens- Input
US$/1M tokens
- Output
US$/1M tokens
– Cache Hit Blended
MiniMax MiniMax-M3 (0, 512K]
MiniMax MiniMax-M3 (512K, 1M]
DeepSeek DeepSeek V4 Pro <
Knowledge Atlas
China
Moonshot
Kimi
Qwen Alibaba Max
Xiaomi MiMO -Pro <
ByteDance -pro
OpenAI
Anthropic
Overseas
Google
Claude Opus
Gemini Pro Preview
Grok Grok
Source: Company data, CMBIGM
Note: data are as of 8 Jun 2026
Minimax: financial forecast and valuation
Driven by the strong demand for AI Native Products and Minimax’s multi-modal capabilities, we
expect Minimax’s total revenue to grow by 260%/161%/135% YoY to US$285/743/1,748mn in
FY26E/27E/28E. AI Native Products/Open Platform and Other AI- based Enterprise Services will
each account for 45%/55% of total revenue in FY28E, vs. 67%/33% in FY25.
For the AI Native Products, we forecast its revenue to increase by 164%/148%/127% YoY to
US$140/347/788mn in FY26/27/28E (49%/47%/45% of total revenue), primarily driven by the
robust revenue growth of Hailuo AI and Minimax agent applications.
For the Open Platform and Other AI-based Enterprise Services, we expect its total revenue to grow
by 458%/173%/143% YoY to US$145/396/960mn in FY26/27/28E (51%/53%/55% of total
revenue), mainly fuelled by the rapid growth of token consumption of M series models and the
Coding Plan. The average daily token consumption of M2 series text models has grown to over six
times in Feb 2026, compared to that of Dec 2025, with token consumption from Coding Plan up by
over ten times.
Figure 26: Minimax: revenue trend
(US$mn) 2,000
1,800
1,600
1,400
1,200
1,000
800
600
400
200
0
2023 2024 2025 2026E 2027E 2028E
300%
250%
200%
150%
100%
50%
0%
Figure 27: Minimax: revenue mix
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Open Platform and Other AI-based Enterprise Services AI
Native Products
YoY (RHL)
2023 2024 2025 2026E 2027E 2028E
Open Platform and Other AI-based Enterprise Services AI
Native Products
Source: Company data, CMBIGM estimates Source: Company data, CMBIGM estimates
Figure 28: Minimax: key financials
(US$ mn) 2024 2025 2026E 2027E 2028E
Total revenue 1,
YoY % % % % %
AI native products
YoY % % % % %
Open platform and other AI-based enterprise services
YoY % % % % %
GPM % % % % %
S&M expense ratio 285% 66% 30% 15% 8%
R&D expense ratio 619% 320% 170% 95% 55%
G&A expense ratio 47% 47% 24% 11% 5%
Adjusted net income -244 -251 -423 -479 -324
Adjusted net margin % % % % %
Source: Company data, CMBIGM estimates
Operating efficiency and margin analysis
We estimate overall GPM to expand from % in FY25 to %/%/% in FY26/27/28E,
mainly attributable to the revenue mix shift to higher-margin Open Platform business, economies
of scale and the improving pricing power of Minimax models.
Figure 29: Minimax: GPM trend
70%
60%
50%
40%
30%
20%
10%
0%
-10%
-20%
Overall GPM
GPM - AI Native Products
GPM - Open Platform and Other AI-based Enterprise Services
Source: Company data, CMBIGM
Despite continuous R&D investment, we expect R&D expenses ratio to decline from 320% in
FY25 to 170%/95%/55% in FY26/27/28E, thanks to the operating leverage. We expect S&M
expenses ratio to drop from 66% in FY25 to 30%/15%/8% in FY26/27/28E, as the company
executes product-lead strategy and reduce investment in user acquisition. We also model G&A
expenses to decline from 47% in FY25 to 24%/11%/5% in FY26/27/28E, supported by the
improved efficiency.
Supported by the operating leverage and enhanced efficiency, we expect adjusted net margin to
improve from -317% in FY25 to -148%/-64%/-19% in FY26/27/28E. Overall adjusted net loss will
widen from US$251mn in FY25, to US$423/479/324mn in FY26/27/28E.
Figure 30: Minimax: margin trend Figure 31: Minimax: operating expense ratio
0%
-100%
-200%
-300%
-400%
-500%
-600%
-700%
-800%
-900%
700%
600%
500%
400%
300%
200%
100%
0%
2024 2025 2026E 2027E 2028E
OPM Adj. NPM R&D% S&M% G&A%
Source: Company data, CMBIGM estimates Source: Company data, CMBIGM estimates
2024 2025 2026E 2027E 2028E
2024 2025 2026E 2027E 2028E
Companies Ticker Price Revenue growth (yoy%) Current PS (x) Revenue CAGR
Figure 32: Minimax: adj. net profit
Source: Company data, CMBIGM estimates
Valuation
We value Minimax at HK$ per share (US$) based on 80x FY26E PS. Our target PS
multiple is at a premium to the global peers, mainly reflecting: 1) Minimax’s robust revenue growth
outlook (25-28E CAGR of 181%) underpinned by the booming AI demand;
2) Minimax’s strong multi-modal capabilities and inference cost efficiency, which help Minimax
establish solid competitive moats.
Figure 33: Minimax: target valuation
Total revenue 285
Target 2026E PS
Target equity valuation 22,794
Valuation per share (US$)
Valuation per share (HK$)
Source: Company data, CMBIGM estimates
Figure 34: Global AI peers: valuation comparison
(Local) 2026E 2027E 2026E 2027E 25-27E
Zhipu 2513 HK 1, 352% 163% 203%
Xunce 3317 HK 89% 66% 68%
Palantir PLTR US 73% 45% 54%
Average
Source: Bloomberg, CMBIGM Note:
data as of 8 Jun 2026
(US$ mn)
0
2023 2024 2025 2026E 2027E 2028E
0%
(100) -50%
(200)
-100%
-150%
(300)
-200%
(400)
-250%
(500) -300%
(600) -350%
Adj. net profit Adj. NPM
P/S Valuation (US$mn) FY26E
Prior experiencePositionName
Appendix
Figure 35: Minimax: shareholding structure (Jan 2026)
Source: Company data, CMBIGM
Figure 36: Management profile
Dr. Yan Junjie Founder, CEO and CTO
Minimax’s founder, the chairman of the Board, executive director, chief
executive officer and chief technology officer. He is primarily responsible for
overseeing the overall management and business operation, board affairs,
financial affairs, formulating strategies and operation plans particularly on AI
research and development, making major business decisions of the company.
Prior to founding the company, Dr. Yan served at SenseTime Group for more
than six years with positions such as its vice president, and vice-head of its
research institute.
Minimax’s executive director and chief operating officer. Ms. Yun is
primarily responsible for overseeing the overall management and business
operation, board affairs, formulating strategies and operation plans
particularly on product and commercialization, making major business
decisions of the company. Prior to joining the company, she
Ms. Yun Yeyi COO
Large language
served various positions in SenseTime Group Inc. She was the manager of
fundraising and strategic investment of SenseTime Group Inc. from
September 2017 to August 2018. She was then promoted to the CEO
executive assistant and head of strategy from August 2018 to January 2021.
Ms. Yun then served as its director of innovative business from January 2021
to January 2022.
Minimax’s large language model research and engineering leader. Mr. Zhao
joined the company as a natural language processing researcher and engineer
since August 2023. He is primarily responsible for the research and
development of large language models. Prior to joining
Mr. Zhao Pengyu model research and
engineering leader
Minimax, Mr. Zhao served as a research software development engineer in
Beijing Hulu Technology Co., Ltd., a company mainly engaged in the
research and development of streaming media technology, between August
2020 and July 2023, where he was primarily responsible for recommendation
algorithms.
Mr. Zhou Yucong
Visual model
research and
engineering leader
Minimax’s visual model research and engineering leader. Mr. Zhou joined
Minimax as a visual model researcher and engineer since March 2022. He is
primarily responsible for research and development of visual models. Prior to
joining Minimax, Mr. Zhou worked at SenseTime Group Inc. from April 2018
to July 2019 and Huawei Technology Company Limited from August 2019 to
March 2022, where he focused on algorithms. Mr. Zhou has also been the
legal representative and director of Shanghai MiniMax since January 2023.
Source: Company data, CMBIGM
Investment risks
Fierce competition in the LLM industry
The LLM industry is highly competitive. Minimax not only competes against global leading
technology companies with strong financial resources and tech infrastructure, but also competes
against AI Labs with high talent intensity and agile organisational structure. The company needs to
continuously innovate and invest to maintain its competitive position. Otherwise, the intensifying
competition may result in slower revenue growth, loss of market share and impaired profitability.
Relatively low floating share base
The company has a relatively low floating share base of as of 1 Jun 2026, which
represented % of total share base. Meanwhile, shares (% of total share base) held
by the pre-IPO and cornerstone investors will exit lock-up period on 9 July 2026. The end of lock-up
period may trigger profit-taking actions of certain investors, potentially leading to selling pressure
on the shares, in our view.
Continuous investment in R&D may affect profitability and cash flow
Minimax’s success depends on its continuous investments in LLM and AI applications to meet
customers’ demand. The company’s R&D expenses were US$70/189/253mn in FY23/24/25,
accounting for 2,023/619/320% of total revenue. Due to the investment, the company recorded adj.
net loss of US$89/244/251mn, and operating cash outflow of US$64/258/280mn in FY23/24/25.
As the LLM industry is subject to constant technological advances and intensive competition, the
company expects to continue to invest in R&D, which may adversely affect profitability and cash
flow.
Failure to attract and retain key talents may affect R&D and operation
The market for high-caliber AI talent is highly competitive. The company needs to attract, retain
and motivate key personnels in the form of cash and stock-based compensation. Unexpected
departure of key employees, including senior executives or core technical personnel, could disrupt
the company’s operations, delay product development or strategic initiatives, and result in the loss
of valuable institutional knowledge.
Evolving regulations may adversely affect business operation
Regulations for AI and LLM industry is evolving rapidly. Minimax is subject to a series of laws
and regulations that will continue to evolve in the future, such as Provisions on the Administration
of Deep Synthesis of Internet-Based Information Services, Interim Measures for the Administration
of Generative Artificial Intelligence Services, and Measures for the Identification of AI-Generated
and Synthesized Content. Non-compliance with the regulations may result in warning, fine, and
suspension in operations.
Financial Summary
INCOME STATEMENT 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec (US$ mn)
Revenue 3 31 79 285 743 1,748
Cost of goods sold (4) (27) (59) (183) (459) (1,039)
Gross profit (1) 4 20 102 284 709
Operating expenses (92) (254) (301) (581) (836) (1,128)
Selling expense (23) (87) (52) (85) (111) (140)
Admin expense (8) (14) (37) (68) (78) (87)
R&D expense (70) (189) (253) (484) (706) (962)
Others 9 36 40 57 59 61
Operating profit (92) (250) (281) (479) (552) (418)
Interest expense (0) (1) (1) (1) (1) (1)
Others (177) (214) (1,590) 0 0 0
Pre-tax profit (269) (465) (1,872) (480) (553) (420)
Income tax 0 0 0 0 0 0
After tax profit (269) (465) (1,872) (480) (553) (420)
Net profit (269) (465) (1,872) (480) (553) (420)
Adjusted net profit (89) (244) (251) (423) (479) (324)
BALANCE SHEET 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec (US$ mn)
Current assets 320 806 1,007 1,370 1,284 1,729
Cash & equivalents 206 289 508 819 648 950
Restricted cash 0 27 20 20 20 20
Account receivables 1 7 11 31 65 122
Prepayment 4 13 16 47 98 185
ST bank deposits 92 26 14 14 14 14
Financial assets at FVTPL 16 295 439 439 439 439
Other current assets 0 147 0 0 0 0
Non-current assets 4 105 81 84 93 113
PP&E 1 1 2 5 13 32
Right-of-use assets 3 3 2 3 3 3
Financial assets at FVTPL 0 95 70 70 70 70
Other non-current assets 0 5 7 7 7 7
Total assets 324 911 1,088 1,454 1,377 1,842
Current liabilities 663 1,708 3,734 335 736 1,525
Short-term borrowings 0 19 35 35 35 35
Account payables 17 51 58 179 449 1,017
Other current liabilities 644 1,633 3,632 98 205 386
Lease liabilities 1 2 1 1 1 2
Contract liabilities 1 2 8 22 45 85
Non-current liabilities 3 2 3 3 3 4
Obligations under finance leases 2 1 1 1 1 1
Other non-current liabilities 1 1 2 2 2 2
Total liabilities 666 1,710 3,737 339 740 1,528
Share capital 0 0 0 4,244 4,318 4,414
Retained earnings (342) (799) (2,648) (3,128) (3,681) (4,101)
Total shareholders equity (342) (799) (2,648) 1,116 637 313
Total equity and liabilities 324 911 1,088 1,454 1,377 1,842
CASH FLOW 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec (US$ mn)
Operating
Profit before taxation (269) (465) (1,872) (480) (553) (420)
Depreciation & amortization 1 2 3 1 3 8
Change in working capital 26 14 6 148 316 645
Others 178 191 1,583 58 75 98
Net cash from operations (64) (258) (280) (272) (158) 331
Investing
Capital expenditure (1) (1) (1) (4) (12) (27)
Net proceeds from disposal of short-term
investments 51 (359) (91) 0 0 0
Others (90) (72) 164 0 0 0
Net cash from investing (40) (431) 72 (4) (12) (27)
Financing
Net borrowings 307 773 428 0 0 0
Proceeds from share issues 0 0 0 589 0 0
Others (1) (2) (4) (1) (1) (1)
Net cash from financing 306 771 423 588 (1) (1)
Net change in cash
Cash at the beginning of the year 5 206 289 508 819 648
Exchange difference 0 1 3 0 0 0
Cash at the end of the year 206 289 508 819 648 950
GROWTH 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec
Revenue na % % % % %
Gross profit na na % % % %
PROFITABILITY 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec
Gross profit margin (%) % % % % %
Operating margin (2,%) (%) (%) (%) (%) (%)
Adj. net profit margin (2,%) (%) (%) (%) (%) (%)
Return on equity (ROE) na na na na (%) (%)
GEARING/LIQUIDITY/ACTIVITIES 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec
Current ratio (x)
Receivable turnover days
Inventory turnover days
Payable turnover days
VALUATION 2023A 2024A 2025A 2026E 2027E 2028E
YE 31 Dec
P/E na ns ns ns ns ns
Source: Company data, CMBIGM estimates. Note: The calculation of net cash includes financial assets.
Disclosures & Disclaimers
Analyst Certification
The research analyst who is primary responsible for the content of this research report, in whole or in part, certifies that with respect to the securities or issuer that the analyst
covered in this report: (1) all of the views expressed accurately reflect his or her personal views about the subject securit ies or issuer; and (2) no part of his or her
compensation was, is, or will be, directly or indirectly, related to the specific views expressed by that analyst in this report.
Besides, the analyst confirms that neither the analyst nor his/her associates (as defined in the code of conduct issued by The Hong Kong Securities and Futures Commission) (1)
have dealt in or traded in the stock(s) covered in this research report within 30 calendar days prior to the date of issue of this report; (2) will deal in or trade in the stock(s)
covered in this research report 3 business days after the date of issue of this report; (3) serve as an officer of any of the Hong Kong listed companies covered in this report;
and (4) have any financial interests in the Hong Kong listed companies covered in this report.
CMBIGM or its affiliate(s) have investment banking relationship with the issuers covered in this report in preceding 12 months.
CMBIGM Ratings
BUY : Stock with potential return of over 15% over next 12 months
HOLD : Stock with potential return of +15% to -10% over next 12 months
SELL : Stock with potential loss of over 10% over next 12 months
NOT RATED : Stock is not rated by CMBIGM
OUTPERFORM : Industry expected to outperform the relevant broad market benchmark over next 12 months MARKET-
PERFORM : Industry expected to perform in-line with the relevant broad market benchmark over next 12 months
UNDERPERFORM : Industry expected to underperform the relevant broad market benchmark over next 12 months
CMB International Global Markets Limited
Address: 45/F, Champion Tower, 3 Garden Road, Hong Kong, Tel: (852) 3900 0888 Fax: (852) 3900 0800
CMB International Global Markets Limited (“CMBIGM”) is a wholly owned subsidiary of CMB International Capital Corporation Limited (a wholly owned
subsidiary of China Merchants Bank)
Important Disclosures
There are risks involved in transacting in any securities. The information contained in this report may not be suitable for the purposes of all investors. CMBIGM does not
provide individually tailored investment advice. This report has been prepared without regard to the individual investment objectives, financial position or special requirements.
Past performance has no indication of future performance, and actual events may differ materially from that which is contained in the report. The value of, and returns from,
any investments are uncertain and are not guaranteed and may fluctuate as a result of their dependence on the performance of underlying assets or other variable market
factors. CMBIGM recommends that investors should independently evaluate particular investments and strategies, and encourages investors to consult with a professional
financial advisor in order to make their own investment decisions.
This report or any information contained herein, have been prepared by the CMBIGM, solely for the purpose of supplying information to the clients of CMBIGM or its
affiliate(s) to whom it is distributed. This report is not and should not be construed as an offer or solicitation to buy or sell any security or any interest in securities or enter
into any transaction. Neither CMBIGM nor any of its affiliates, shareholders, agents, consultants, directors, officers or employees shall be liable for any loss, damage or
expense whatsoever, whether direct or consequential, incurred in relying on the information contained in this report. Anyone making use of the information contained in this
report does so entirely at their own risk.
The information and contents contained in this report are based on the analyses and interpretations of information believed to be publicly available and reliable. CMBIGM has
exerted every effort in its capacity to ensure, but not to guarantee, their accuracy, completeness, timeliness or correctness. CMBIGM provides the information, advices and
forecasts on an "AS IS" basis. The information and contents are subject to change without notic e. CMBIGM may issue other publications having information and/ or
conclusions different from this report. These publications reflect different assumption, point-of-view and analytical methods when compiling. CMBIGM may make
investment decisions or take proprietary positions that are inconsistent with the recommendations or views in this report.
CMBIGM may have a position, make markets or act as principal or engage in transactions in securities of companies referred to in this report for itself and/or on behalf of its
clients from time to time. Investors should assume that CMBIGM does or seeks to have investment banking or other business relationships with the companies in this report.
As a result, recipients should be aware that CMBIGM may have a conflict of interest that could affect the objectivity of this report and CMBIGM will not assume any
responsibility in respect thereof. This report is for the use of intended recipients only and this publication, may not be reproduced, reprinted, sold, redistributed or published
in whole or in part for any purpose without prior written consent of CMBIGM.
Additional information on recommended securities is available upon request.
For recipients of this document in the United Kingdom
This report has been provided only to persons (I)falling within Article 19(5) of the Financial Services and Markets Act 2000 (Financial Promotion) Order 2005 (as amended
from time to time)(“The Order”) or (II) are persons falling within Article 49(2) (a) to (d) (“High Net Worth Companies, Unincorporated Associations, etc.,) of the Order, and
may not be provided to any other person without the prior written consent of CMBIGM.
For recipients of this document in the United States
CMBIGM is not a registered broker-dealer in the United States. As a result, CMBIGM is not subject to . rules regarding the preparation of research reports and the
independence of research analysts. The research analyst who is primary responsible for the content of this research report is not registered or qualified as a research analyst with
the Financial Industry Regulatory Authority (“FINRA”). The analyst is not subject to applicable restrictions under FINRA Rules intended to ensure that the analyst is not
affected by potential conflicts of interest that could bear upon the reliability of the research report. This report is intended for distribution in the United States solely to "major
US institutional investors", as defined in Rule 15a-6 under the US, Securities Exchange Act of 1934, as amended, and may not be furnished to any other person in the United
States. Each major US institutional investor that receives a copy of this report by its acceptance hereof represents and agrees that it shall not distribute or provide this report
to any other person. Any . recipient of this report wishing to effect any transaction to buy or sell securities based on the information provided in this report should do so
only through a .-registered broker-dealer.
For recipients of this document in Singapore
This report is distributed in Singapore by CMBI (Singapore) Pte. Limited (CMBISG) (Company Regn. No. 201731928D), an Exempt Financial Adviser as defined in the
Financial Advisers Act (Cap. 110) of Singapore and regulated by the Monetary Authority of Singapore. CMBISG may distribute reports produced by its respective foreign
entities, affiliates or other foreign research houses pursuant to an arrangement under Regulation 32C of the Financial Advisers Regulations. Where the report is distributed in
Singapore to a person who is not an Accredited Investor, Expert Investor or an Institutional Investor, as defined in the Securities and Futures Act (Cap. 289) of Singapore,
CMBISG accepts legal responsibility for the contents of the report to such persons only to the extent required by law. Singapore recipients should contact CMBISG at +65
6350 4400 for matters arising from, or in connection with the report.
