M sight
Asia Technology
China AI – Entering a New
Phase
4 20 7677-1018
+852 2239-7357
4 20 7425-2915
+852 2848-6511
+852 3963-4015
+852 2848-6918
86 2 2730-1725
86 2 2730-2865
+852 2239-1911
+852 2848-7310
+852 2239-1572
tive
China is on a path to drive simultaneous AI advancements
frontier sectors, spanning robotics, new-energy vehicles, a
LLMs. Its "AI Plus" initiative is rapidly injecting its newfound
capabilities and AI agents into world-class smart products,
services and embodied intelligence.
China’s AI is entering a new phase – one de
less by catching up in capability and more
capturing value. The narrative has shifted d
from training to inference, from technology
application, and from potential to real earnin
Rather than competing purely at the frontier
is optimizing for speed, cost efficiency, and s
level integration, enabling AI to diffuse rapid
the real economy. At the same time, China
semiconductor self-sufficiency rate continues to improve from 41% in 2025
by 2030 in our estimates, supporting a more resilient and cost-efficient dep
At the macro level, AI will become a medium-term productivity lever. We
estimate that AI could lift China's total factor productivity growth by ~3ppt
cumulatively over the next decade, partly offsetting the aging population an
bringing 2035 potential GDP to a level higher than the trajectory wit
adoption. But its near-term growth impact will likely be muted, with positiv
AI-related capex cycle and modest initial productivity gains offset by transit
labor disruption.
Capturing the "rate of change". AI adoption is broadening, with enabler/ad
penetration rising to ~51% of coverage from 43% in two years. Our latest 1H
AlphaWise China CIO survey indicates that 47% plan to start their first AI pr
within the next year (an increase on 40% in 2H25). The economics are becom
tangible: AI adopters have seen NTM EPS rise by 62% in the past two years,
significantly outperforming MSCI China at 10%, alongside 12-13ppt EBIT mar
expansion to 16-17% by 2027E, well ahead of the broader market.
Investment Implications. While enablers and foundational models remain t
investment thesis at the current stage, widespread adoption brings potentia
investment opportunities to those who reap the fruits of AI.
• Beisen (), Meitu (), Beijing Roborock (
Group () and Ecovacs Robotics () stand out in
risk-reward analysis among AI adopters.
• MiniMax () and () are the key foundational mode
providers in China AI, while Alibaba () is the best-positioned fu
th
s a result,
a conflict of
Stanley
ley
vestment
closures,
d of this
t registered
May 10, 2026 09:00 PM GMT
ct
g
key
idea
our
Morgan Stanley does and seeks to do business wi
companies covered in Morgan Stanley Research. A
investors should be aware that the firm may have
interest that could affect the objectivity of Morgan
Research. Investors should consider Morgan Stan
Research as only a single factor in making their in
decision.
For analyst certification and other important dis
refer to the Disclosure Section, located at the en
report.
+= Analysts employed by . affiliates are no
Asia Pacific In
Morgan Stanley & Co. International plc+
Shawn Kim
Equity Analyst
@ +4
Morgan Stanley Asia Limited+
Duan Liu
Equity Analyst
@
Morgan Stanley & Co. International plc+
Cindy Huang
Equity Analyst
@ +4
Morgan Stanley Asia Limited+
Robin Xing
Chief China Economist
@
Jenny Zheng, CFA
Economist
@
Gary Yu
Equity Analyst
@
Morgan Stanley Taiwan Limited+
Charlie Chan
Equity Analyst
@ +8
Sharon Shih
Equity Analyst
@ +8
Morgan Stanley Asia Limited+
Yang Liu
Equity Analyst
@
Daisy Dai, CFA
Equity Analyst
@
Lydia Lin
Equity Analyst
@
S. Korea Technology
Asia Pacific
Industry View Attrac
across
nd
AI
fined
by
ecisively
to
gs.
, China
ystem-
ly across
to 86%
loyment.
d
hout AI
es from
ional
opter
26
oje
in
gin
he
l
), M
l
ll-
with FINRA, may not be associated persons of the member
and may not be subject to FINRA restrictions on
communications with a subject company, public appearances
and trading securities held by a research analyst account.
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
mailto:@
M Asia Pacific Insight
2
stack AI platform in our coverage.
• We highlight CATL ( Yingliu () and
Sieyuan () as the key names in power
• We continue to like Cambricon (), Iluvatar (), NAURA
(), AMEC (), ACMR (), SMIC () and
Unimicron () as the key AI enablers that benefit from China
Semiconductor localization secular trend.
ic Insight
M Asia Pacif
Other Contributors
Morgan Stanl 3
M Asia Pacific Insight
4
Executive Summary
China’s AI story is no longer about catching up – it is about rewriting the rules of the
game.
Over the past year, the market narrative has shifted decisively. China was once seen as
a structurally constrained AI ecosystem – limited by access to leading-edge chips and
frontier research – but it has since demonstrated a different kind of strength: the ability to
scale, adapt, and commercialize AI faster than anywhere else. This divergence is not a
weakness. It is China’s defining edge.
As we highlighted in last year’s Blue Paper (China AI: The Sleeping Giant Awakens), China’s
AI model is fundamentally different — state-coordinated, data-rich, and industrially
integrated. Today, that model is moving from framework to execution. AI is no longer
confined to labs or hyperscaler capex cycles but it is spreading rapidly across the real
economy, embedding itself into energy systems, manufacturing processes, consumer
platforms, and urban infrastructure.
A Shift in the Center of Gravity. The global AI race is often framed around model
supremacy, but China is playing a different game. Rather than chasing the absolute
frontier, China is optimizing for speed of deployment, cost efficiency, and system-level
integration. As AI development is moving from training to inference and applications,
three key trends have emerged in the past 12 months.
• AI has moved from infrastructure to earnings. China AI adopters are now
achieving tangible financial results – rising earnings expectations and structurally
expanding margins. AI is driving efficiency gains and operating leverage, not just
revenue growth, reinforcing its role as a productivity engine.
• The bottleneck has shifted — from compute to power and deployment. The
constraint is no longer whether AI can be built, but whether it can be powered,
scaled, and delivered in real time. This is driving a new investment cycle in energy
systems, particularly storage and grid flexibility, as AI infrastructure collides with
physical-world limitations. The exponential demand is colliding with real-world
power constraints, making energy storage and grid intelligence critical enablers.
• The new battlegrounds are moving to new energy systems and physical AI.
From humanoid robots to autonomous vehicles, AI is becoming embodied. China’s
advantage — its manufacturing base, supply chain depth, and real-world data —
positions it uniquely to lead in this next phase of “physical AI.”
What Has Changed?
China has made meaningful progress in several foundational areas:
• AI commercialization at scale: Leveraging its large domestic market and super-app
ecosystem, China is accelerating adoption across both consumer (2C) and
enterprise (2B) use cases, particularly via low-cost, open-source models and
integrated platforms.
• Infrastructure and ecosystem buildout: Hyperscaler capex, data center expansion,
and AI cloud growth (72% CAGR expected through 2029) are enabling large-scale
M Asia Pacific Insight
Morgan Stanl 5
deployment and inference growth.
• More self-sufficient: Domestic AI chips and system-level optimization are
narrowing the compute gap, with self-sufficiency improving and competitiveness
increasingly evaluated on cost-performance rather than leading-edge nodes.
What Still Needs Developing?
Despite these advances, several structural constraints remain:
• Compute and semiconductor bottlenecks: Limited access to cutting-edge nodes,
EDA tools, and advanced manufacturing continues to constrain frontier innovation.
• Monetization lag in enterprise AI: While adoption is accelerating, especially in
cost-efficiency use cases, revenue generation — particularly in 2B applications —
remains nascent and uneven.
• Labor and macro adjustment risks: AI’s near-term impact is likely to be disruptive,
with potential labor displacement preceding productivity gains, posing challenges
for income growth and demand stability.
Macro Implications
AI raises China’s potential growth, but as a medium‑term productivity story rather
than a near‑term cyclical boost: We estimate that AI could lift China’s total factor
productivity growth by around 3ppt cumulatively over the next decade, leaving the level
of potential GDP roughly higher by 2035 than under a no‑AI counterfactual. In the
near term, however, the contribution to headline growth is likely to be muted: while an
AI‑driven capex upcycle and early task‑level efficiency gains provide support, these are
largely offset by transitional labour disruptions.
Indeed, despite lower aggregate AI job exposure than in the US, China’s rapid diffusion
pace and starting point of weak corporate earnings mean displacement effects could
materialise faster than job creation over the next 2–3 years. This risk would be amplified
by the broad‑based nature of AI substitution, spanning junior white‑collar roles
(exacerbating youth unemployment), mid‑level professional jobs (adding pressure to the
already stressed middle-income group), and even labour‑intensive services (the main
absorber of employment losses in recent years). This would add to deflationary pressures
if not addressed properly.
Policy mitigation is thus expected to intensify. But protection is likely to skew toward
labour‑intensive services, while higher‑end cognitive roles face greater adjustment to
preserve GenAI competitiveness — implying only a partial offset to near‑term growth
drags.
Investment Implications
AI “rate of change” underpriced by the market. Over the past 12 months, AI enablers —
particularly power, semiconductors and infrastructure — have continued to outperform,
driven by the initial capex cycle and strong revenue visibility across the supply chain.
However, we believe the next phase of the China AI cycle will be characterized by a
meaningful catch-up in application-layer performance, as AI adoption broadens and begins
M Asia Pacific Insight
6
to translate into tangible incremental earnings and margin expansion.
AI adoption accelerating. According to our China CIO survey, AI/Machine Learning/
Process Automation demand stayed strong and edged closer to the peak in 1H25 with
biggest biannual increase in spending expectations. Expectations for AI/LLM-related IT
investments entering production in 2026 have decreased slightly but remain widespread.
Currently, 27% of organizations have launched their first AI/LLM project (a decline from
30% in 2H25), while 47% plan to start within the next year (an increase on 40% in 2H25),
and 15% expect to begin in two years or later (down from 20% in 2H25).
AI adopters have seen NTM EPS rise ~60% since late 2023 and EBIT margins expand by
12–13ppt through 2027E, significantly outperforming the broader market. Beisen, Meitu,
Beijing Roborock, Midea Group, Ecovacs Robotics stand out on risk-reward among AI
adopter stocks (more details: China AI Adopter).
Across the stack, Minimax and are key foundational model provides in China AI, and
we continue to see structural leadership among China’s internet platforms. We view
Alibaba as the best-positioned full-stack AI platform, while Tencent remains strongest in
the application layer, leveraging its ecosystem to drive monetization and user
engagement. Cambricon, Iluvatar, Naura, AMEC, ACMR, SMIC continue to be our key
picks under semi localization themes. We like Unimicron, Shennan Circuit, Shengyi Tech
and Zhen Ding as well for their substrate/PCB/CCL exposure to AI servers using local AI
chips; server OEM/ODM like Inspur and Lenovo also stand out to us as beneficiaries of an
increasingly localized AI server supply chain, alongside Delta for its AI server power
management/cooling solutions. We also highlight CATL, Yingliu and Sieyuan as the key
names in power.
Exhibit 1: China AI 65
Source: Morgan Stanley Research; Red box highlights the newly added names compared to previous version
M Asia Pacific Insight
Morgan Stanl 7
Exhibit 2: China AI 65 Share Price Performance comparison
50
100
150
200
250
300
350
May-25 Jun-25 Jul-25 Aug-25 Sep-25 Oct-25 Nov-25 Dec-25 Jan-26 Feb-26 Mar-26 Apr-26 May-26
Past 12 Months Rebased Share Performance
Power Semi & Hardware Infrastructure Foundational Models Applications MSCI China
Source: FactSet, Morgan Stanley Research
Methodology: Constructing the “China AI 65” Universe
The “China AI 65” framework is designed to identify companies with meaningful
and monetizable exposure to AI across the full value chain. Our methodology
combines a top-down thematic screening process with bottom-up analyst
validation and quality control to ensure robustness. This approach builds on our
proprietary Global AI Mapping Survey and sector research, which track both the
degree of AI exposure and its financial materiality across our Greater China
coverage universe.
Screening
Leveraging our Global AI Mapping Survey, we construct an initial universe based on
AI exposure and categorization (enabler, adopter, wildcard etc) in Greater China
coverage universe. Then, rather than focusing solely on static levels of AI exposure,
we prioritize companies where AI relevance is increasing compared to the survey
last year — either through strengthening strategic positioning and narrative, or
material EPS impact for the next 12 months.
Details of the screening results: China AI Survey – AI Impact Broadening Out
Bottom-up Validation and Quality Control
Bottom-up analyst input is then applied to validate each name. Sector teams
assess how AI translates into revenue growth, cost efficiency, or margin expansion,
supported by product integration, customer adoption, and competitive positioning
(., data, ecosystem, or scale advantages). We rely on analysts research and
judgement to avoid "AI washing". Names with purely narrative-driven exposure or
without a credible monetization pathway are excluded.
Details of individual stock narrative: China AI Adopter
Risks and Limitations
• AI monetization — particularly in enterprise use cases — remains early and
uneven, and rapid technological change may shift competitive positioning
over time.
• Data transparency varies across sectors, and the framework may favor
companies with more visible near-term impact, potentially underweighting
M Asia Pacific Insight
8
earlier-stage opportunities.
• Outcomes also remain sensitive to macro and policy factors, including
semiconductor constraints and infrastructure availability.
Performance
To analyze equity market performance across the AI ecosystem, we classify China
AI 65 companies into five thematic segments spanning the AI value chain. Thematic
basket performance is calculated using fixed point-in-time constituents as of May
2025. Constituents remain unchanged throughout the analysis period, with no
subsequent additions, deletions, or category migration incorporated into returns.
We adopt this methodology to maintain consistency and comparability across
performance series over time.
M Asia Pacific Insight
Morgan Stanl 9
China AI Survey – AI Impact Broadening Out
Our fifth Global AI Mapping Survey shows broadening AI adoption across the MS
coverage universe while the latest thematic ranking shows China's AI path retaining top
rank, given earnings prospects and valuations.
In this section, we leverage the global survey data to dive into the latest trends in China AI
adoption.
Relevant link: Asia Thematic Alpha: Mapping AI Adoption and Disruption Opportunities
(February 11, 2026)
Steady Growth in AI Adoption
China markets continue to see AI adoption broadening out, with the share of Adopters and
Enabler/Adopters, as identified by our analysts, rising from 31% in our inaugural survey
(two years ago) to 37% in the latest survey. The materiality of AI impact has also
increased: the share of companies who faces "Core to thesis" and "Significant" impact from
AI increased from 11% to 16% over the last two years, with at least 37% of the companies
now see moderate AI impact vs. 32% two years ago.
Exhibit 3: AI adoption broadening – enabler/adopter increased
from 43% to 51% in 2 years
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Jan-24 Jun-24 Jan-25 May-25 Jan-26
S
h
a
re
o
f
m
a
p
p
e
d
n
a
m
e
s
Survey wave
Exposure distribution across waves
Enabler/Adopter Enabler Adopter Protected Wildcard Disrupted Don't Know
Source: Morgan Stanley Research
Exhibit 4: 16% of the companies saw significant AI impact
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Jan-24 Jun-24 Jan-25 May-25 Jan-26
S
h
a
re
o
f
m
a
p
p
e
d
n
a
m
e
s
Survey wave
Materiality distribution across waves
Core to Thesis Significant Moderate Insignificant Don't Know
Source: Morgan Stanley Research
AI Adopters To Benefit From Cost Efficiency First
Similar to what we've seen in the global survey, most AI adopters in China will first benefit
from cost efficiency (91% in China vs. 74-90% in global survey) vs. revenue growth (9% vs.
10-26% in global survey), reinforcing the view that AI initially will have more impact on
margins than on top-line acceleration. Revenue growth at the current stage dominates
among the AI enablers, where 79% of the enablers see the impact, consistent with what
we've seen on the aggressive AI capex that benefit the entire supply chain (IT) as well as
infrastructure companies (utilities, materials). Excluding the enabler concentrated
segments, consumer discretionary (EV, home appliances) and real estate (property
management) see the most significant revenue growth from AI.
M Asia Pacific Insight
10
Exhibit 5: 91% of the adopters benefit from cost efficiency
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Enabler Enabler/Adopter Adopter
S
h
a
re
o
f
g
ro
u
p
r
e
s
p
o
n
s
e
s
Exposure group
Benefit channel by AI exposure
Cost Efficiency Revenue Growth
Source: Morgan Stanley Research
Exhibit 6: IT, Utilities and Materials see the most significant
revenue growth from AI
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Information
Technology
Utilities Materials Communication
Services
Industrials Consumer
Discretionary
Real Estate Energy Financials Health Care Consumer
Staples
S
h
a
re
o
f
s
e
c
to
r
re
s
p
o
n
s
e
s
Sector
Benefit channel by sector
Cost Efficiency Revenue Growth
Source: Morgan Stanley Research
Rate of Change to Drive Outperformance
To assess the relevance of AI's rate of change, we looked at the performance of stocks for
which our China analysts identified a change in materiality and exposure as part of the re-
mapping exercise. We see substantial outperformance for stocks where the rate of change
is positive. Comparing the fifth wave of mapping to the fourth, consumer discretionary
groups show a notable rise in both materiality and exposure. Within this segment, 9% saw
an increase in materiality and 11% an increase in exposure. Within Utilities, 44% saw an
increase in exposure due to increasing relevance of power to data center build out. In
industrials, materiality rose for 16% driven by the humanoid-related supply chain.
Exhibit 7: Rate of change breakdown by segment
Information Technology
Health Care
Materials
Consumer Staples
Financials
Communication Services
Real EstateEnergy
-1%
1%
3%
5%
7%
9%
11%
0% 1% 2% 3% 4% 5%
R
is
in
g
E
x
p
o
s
u
re
%
Rising Materiality %
Rate of change by sector
45%
Utilities
8% 16%
Consumer
Discretionary
Industrials
Source: Morgan Stanley Research
Exhibit 8: Companies with increased AI impact materiality
outperformed MSCI China by 88ppt in the past year
70
90
110
130
150
170
190
210
Mar-25 Apr-25 May-25 Jun-25 Jul-25 Aug-25 Sep-25 Oct-25 Nov-25 Dec-25 Jan-26 Feb-26 Mar-26
Postive rate of change is a performance driver
Increased Materiality Basket MSCI China
Source: Morgan Stanley Research
M Asia Pacific Insight
Morgan Stanl 11
China AI Adopter
Leveraging our Global AI Survey and analyst insights, we've selected 39 names under our
Greater China coverage as the AI adopters that would win through cost efficiency and top-
line/margin growth in the AI era.
AI Adopters See a Forward Earnings Boost
China AI adopters have seen a clear forward earnings up-trend since late 2023, with NTM
EPS rising from (Dec-23 rebased) to around 167 by May-26, materially
outperforming MSCI China, which stayed broadly flat over the same period ( Exhibit 9 ).
This suggests the market has increasingly priced in stronger NTM earnings expectations
for companies with clearer AI adoption stories.
Please refer to the table at the end of this section on how AI drives individual stocks
narrative
Margin Expansion Supports the AI Adoption Case
EBIT margins for China AI adopters have expanded steadily from around 4% in 2021 to 16-
17% by 2027E, implying 12-13ppt of cumulative margin expansion over the period. The
trajectory reinforces our view that, at this stage, AI adoption is translating first into
efficiency gains and operating leverage rather than purely top-line acceleration.
Who are better positioned?
On our risk-reward framework, Beisen Holding, Meitu and Beijing Roborock Technology
screen as the most attractive names, combining the strongest risk-adjusted base return
with favorable skew. More broadly, stocks positioned in the upper-right quadrant in
Exhibit 55 look best placed to outperform, in our view. We also compared 2027e ROE vs.
current valuation relative to past three year average to see if the AI upside potential has
been fully captured by current market expectations, Midea Group and Ecovacs Robotics
stand out in the screening in addition to the names highlighted above. FII, China Resources
Mixc Lifestyle Services, Futu and CATL stand out in ROE return while the valuation
reflects that market has already partially priced in the AI potential.
• Beisen: We like Beisen for its incremental revenue from AI with high profitability
and market share gain, with AI ARR, and we expect it to grow 10x from Rmb6mn+
in FY25 to Rmb60mn+ in FY26.
• Meitu: We expect Meitu to successfully monetize AI through its photo, video, and
design applications with paying ratio to double in three years to 8-10% by 2028.
• Beijing Roborock Technology: Advanced AI algorithms to support premium
positioning, faster replacement cycle, and improving gross margin with future
potential in home-use service robots supported by its robotic cleaners with
mechanical arms and wheel-leg.
• Midea Group: Key AI beneficiary as one of the global leading robot producers
(KUKA) and increase quickly in China. It is transforming from a traditional home
appliances company to a smart manufacturing/intelligent building technology/
energy solution provider. End-to-end AI development (smart appliances + “factory
brain” automation) supports both top-line growth via whole-home intelligence
ecosystems and manufacturing/industrial scenario and applications, which also
M Asia Pacific Insight
12
drives structural margin expansion via highly automated production.
• Ecovacs Robotics: AI powered perception and navigation upgrades enable
production premiumization and TAM expansion.
Exhibit 9: AI Adopters NTM EPS trend outpaced MSCI China
D
e
c
-2
3
J
a
n
-2
4
F
e
b
-2
4
M
a
r-
2
4
A
p
r-
2
4
M
a
y-
2
4
J
u
n
-2
4
J
u
l-
2
4
A
u
g
-2
4
S
e
p
-2
4
O
c
t-
2
4
N
o
v-
2
4
D
e
c
-2
4
J
a
n
-2
5
F
e
b
-2
5
M
a
r-
2
5
A
p
r-
2
5
M
a
y-
2
5
J
u
n
-2
5
J
u
l-
2
5
A
u
g
-2
5
S
e
p
-2
5
O
c
t-
2
5
N
o
v-
2
5
D
e
c
-2
5
J
a
n
-2
6
F
e
b
-2
6
M
a
r-
2
6
A
p
r-
2
6
Consensus Estimates - NTM EPS (2023 Dec rebased = 100)
China AI Adopter MSCI China
Source: Morgan Stanley Research
Exhibit 10: AI Adopters Are Seeing EBIT Margin Expansion
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
2021 2022 2023 2024 2025 2026 2027
EBIT Margin
Source: Morgan Stanley Research
Exhibit 11:
Risk-Reward Skew (Bull+Bear)
Meitu
Bilibili
Baidu
Tencent Holdings
Kuaishou
China Literature
Alibaba Group Holding
Ecovacs Robotics
Beijing Roborock
Midea Group
Haier Smart Home
Yum China
TAL Education Group
Futu Holdings
Bank of Ningbo
Ping An Insurance Group
ZhongAn Online P & C Insurance
Alibaba Health
Information
Technology
JD Health
Dian Diagnostics
Guangzhou Kingmed Diagnostics
Shanghai United Imaging Healthcare
Insilico Medicine
WuXi AppTec
CATL
Beijing Geekplus
ZTO Express
China TransInfo Technology
Advantech
Dahua Technology
BYD Electronics
Xiaomi Corp
HIKVision Digital Technology
China Resources Mixc Lifestyle Services
FII
()
()
-
() () -
Skew (Bull+Bear)/Vol (RV10Y)
Base Return/Vol (RV10Y)
Most Attractive
Least Attractive
Beisen
Source: Morgan Stanley Research
M Asia Pacific Insight
Morgan Stanl 13
Exhibit 12: P/B vs. ROE
Meitu
Bilibili
Baidu
Tencent Holdings
Kuaishou
China Literature
Alibaba Group Holding
Ecovacs Robotics
Beijing Roborock
Midea Group
Haier Smart Home
Yum China
TAL Education Group
Futu Holdings
Bank of Ningbo
Ping An Insurance Group
ZhongAn Online P & C Insurance
Alibaba Health Information Technology
JD Health
Dian Diagnostics
Guangzhou Kingmed Diagnostics
Shanghai United Imaging Healthcare
WuXi AppTec
CATL
Beijing Geekplus
ZTO Express
Beisen Holding
China TransInfo Technology
Advantech
Dahua Technology
BYD Electronics
Xiaomi Corp
HIKVision Digital Technology
Shenzhen Transsion Holdings
0%
5%
10%
15%
20%
25%
30%
35%
40%
-55% -45% -35% -25% -15% -5% 5% 15% 25%
2027e ROE
P/B valuation discount/premium vs. L3Y Avg
FII
China resources mixc Lifestyle services
45%
Least Attractive
Most Attractive
Source: Morgan Stanley Research
Company AI Story Analyst Report
Beisen Holding Limited Beisen is a rare B2B software provider in China that can generate
incremental revenue from AI with high profitability (80-90% gross
margins), as its HCM AI translates to actual IT spending increases by
taking wallet share from human HR personnel and offline training
institutions, with AI ARR expected to grow 10x from Rmb6mn+ in FY25 to
Rmb60mn+ in FY26.
Lin, Lydia Beisen Holding Limited: 1HF26
Results: Beat and Raise (26 Nov
2025)
Meitu Inc Meitu successfully monetizing AI through its photo, video, and design
applications with paying ratio expected to double in 3 years to 8-10% by
2028.
Lin, Lydia Meitu Inc: Is AI Eating Apps? It's
the Last Mile that Matters (2
Sep 2025)
Beijing Roborock
Technology Co Ltd
Advanced AI algorithms (mapping, obstacle avoidance, automation)
support premium positioning and faster replacement cycles, sustaining
revenue growth with improving gross margins. Company has launched
robotic cleaners with mechanical arms and a “wheel-leg” which we see as
key for further developing home-use service robots.
Ling, Hildy Beijing Roborock Technology
Co Ltd: 3Q25 mild profit miss;
4Q operation is still tracking
healthy (30 Oct 2025)
Midea Group Co Ltd. An AI enabler as the one of leading robot producers. Midea‘s automation
business KUKA is one of the leading automation/robotics player globally
and increase quickly in ’s transforming from a traditional home
appliances company to a smart manufacturing/intelligent building
technology/energy solution provider. End-to-end AI development (smart
appliances + “factory brain” automation) supports both top-line growth
via whole-home intelligence ecosystems and manufacturing/industrial
scenario and applications, which also drives structural margin expansion
via highly automated production.
Lou, Lillian Midea Group Co Ltd.: 3Q25
Results: Solid Execution (29 Oct
2025)
Ecovacs Robotics Co
Ltd
AI-powered perception and navigation upgrades enable product
premiumization (higher ASP robotic cleaners) and expand TAM into
service robotics. The company is expanding product categories
horizontally and developing cleaning robots in various scenarios (.
floors, windows, swimming pools, etc).
Ling, Hildy Ecovacs Robotics Co Ltd:
3Q25's solid growth and healthy
momentum should carry over
to 4Q (26 Oct 2025)
M Asia Pacific Insight
14
ZhongAn Online P & C
Insurance Co Ltd
Its digital bank business has shown very strong growth momentum since
it was set up. The company’s Tech exports have maintained solid growth
in both domestic and overseas markets. ZhongAn has deployed its
stablecoins business, including investments in potential stablecoin
issuers RD InnoTech and ZA Bank, which can provide reserve banking
services.
Zhao, Rick ZhongAn Online P & C
Insurance Co Ltd: 1H25 –
Strong Earnings Growth and
Superior CoR; Bank Turned
Profitable (20 Aug 2025)
ZhongAn Online P & C
Insurance Co Ltd: Revaluation
Backed by Healthy P&C and
Strong Tech Presence (11 Jul
2025)
Alibaba Group Holding Alibaba owns the biggest cloud infrastructure in China and globally #4,
pursuing infrastructure-led AI strategy centered on AliCloud. Cloud
infrastructure to AliCloud is key, be it chips, data centers, cloud platforms
or AI models.
Yu, Gary China's Emerging Frontiers:
China's AI Path: Monetizing
Surging Token Use via AI Cloud
(16 Mar 2026)
Bank of Ningbo Co.
Ltd
Bank of Ningbo has been investing ~5% of revenue annually into
technology and R&D (vs 2-3% at other major banks), allowing digital
transformation across financial products, service channels, marketing
and customer management as well as operational risk control. For
example, it aggregates customs data from over 200 countries, cleans and
standardizes the data, and builds database that clients can directly use
for analysis and decision-making. With broader AI adoption under its
“Technology+” strategy, the bank can further improve product
development efficiency and embed more AI-enabled functionalities into
products, which will help enhance customer stickiness and increase the
conversion rate into financial business.
Xu, Richard Bank of Ningbo Co. Ltd: Small
Gem Shines Big Again (17 Mar
2026)
Xiaomi Corp Integrates AI capabilities across smartphone, EVs, and smart home
devices. Rising investment in AI infrastructure and large language model
(LLM) to enhance competitiveness.
Meng, Andy Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
Shanghai United
Imaging Healthcare Co
AI powered imaging software and workflow automation enhance
diagnostic accuracy and recurring software value, strengthening UIH’s
high end modality competitiveness.
Yan, Alexis China Healthcare: AI Diffusion
in China Healthcare – The
DeepSeek Effect (10 Mar 2025)
Tencent Holdings Ltd. Tencent is adopting a more application-driven AI strategy. Rather than
emphasizing infrastructure leadership, Tencent focuses on leveraging the
WeChat ecosystem, staying on the top of the funnel. (Wechat + Yuanbao)
Yu, Gary China's Emerging Frontiers:
China's AI Path: Monetizing
Surging Token Use via AI Cloud
(16 Mar 2026)
Haier Smart Home Co
Ltd
Vertical AI models and ecosystem integration enable recurring revenue
from smart home services and potential in-home robotics, while
enhancing product ASP through personalized experiences.
Lou, Lillian Haier Smart Home Co Ltd:
3Q25 Results – Growth in Line
(30 Oct 2025)
Ping An Insurance
Group Co of China Ltd
The company focuses on cutting edge applications and has built full
stack AI capabilities. High quality training data and proprietary training
algorithms and code provide clear advantages versus peers. P&C stands
to benefit most from AI and refined management practices, while other
subsidiaries should also gain from higher productivity, stronger risk
control, better claims handling, greater operational efficiency, and
improved sales support over time.
Xu, Richard Ping An Insurance Group Co of
China Ltd: Trip Takeaways -
Improving ROE on Solid Group
Strategy and Synergies; AI
Impresses in Favorable
Environment (1 Feb 2026)
Ping An Insurance Group Co of
China Ltd: Reaching New
Pinnacles (4 Dec 2025)
M Asia Pacific Insight
Morgan Stanl 15
BYD Electronics Expanded into AI data center and related components businesses. Meng, Andy Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
China Literature Ltd China Lit is on its way to further monetize its IP library in short drama and
short animation drama. We believe the ongoing upgrading AI tech would
help China Lit to produce such content at a faster pace and with lower
cost. Hence, we believe China Lit is one of the AI adopters in China.
Xu, Rebecca China Literature Ltd: Revenue
Growth Driven by AI and IP
Strength (17 Mar 2026)
Foxconn Industrial
Internet Co. Ltd.
FII uses digital twins integrating NVIDIA Omniverse libraries to design,
deploy, and manage complex, high-volume production facilities, including
those for AI server systems production.
Shih, Sharon Foxconn Industrial Internet Co.
Ltd.: AI Exposure Also Drives
Margin Expansion; Price Target
Up to Rmb82 (13 Nov 2025)
TAL Education Group TAL leverages AI to transition from traditional tutoring to an integrated
"Content x AI x Software & Hardware" model, aimed at delivering truly
individualized teaching that was previously unscalable
Wang, Eddy TAL Education Group: Strong
F3Q26 Margin Should Trigger
Further Re-rating (30 Jan 2026)
JD Health International
Inc.
AI supports O2O healthcare logistics, online consultation triage and
operational efficiency, with ROI balanced against upfront investment
intensity.
Yan, Alexis China Healthcare: AI Diffusion
in China Healthcare (16 Jul
2024)
Dahua Technology Co.
Ltd.
Deploys Xinghan large-scale AI models to power video-centric AIoT
solutions for intelligent surveillance, traffic management, and smart
cities.
Meng, Andy Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
Guangzhou Kingmed
Diagnostics
AI and LLM integration optimize lab workflows, automate test
interpretation and enable scalable patient follow up, creating cost
efficiencies and future data monetization optionality.
Yan, Alexis China Healthcare: AI Diffusion
in China Healthcare – The
DeepSeek Effect (10 Mar 2025)
Alibaba Health
Information
Technology
AI improves traffic monetization, merchant marketing efficiency and
clinical service matching, leveraging Alibaba’s LLM and data ecosystem.
Yan, Alexis China Healthcare: AI Diffusion
in China Healthcare (16 Jul
2024)
Shenzhen Transsion
Holdings Co Ltd
Partnership with leading AI companies to offer AI features in mid-to-low-
end smartphones.
Meng, Andy Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
Baidu Inc Baidu has been most aggressive in transforming core business into AI-
driven services as a solution to AI disruption of search. It utilizes the Ernie
LLM through search ads transformation, AI cloud infra, and Apollo Go.
Yu, Gary China's Emerging Frontiers:
China's AI Path: Monetizing
Surging Token Use via AI Cloud
(16 Mar 2026)
Bilibili Inc Bilibili will benefit from: 1) increasing advertising budget from the AI
model developers; 2) internal adoption of AI to enhance advertising
efficiencies and drive higher ECPM; and 3) developing its own AI tools for
video makers.
Liu, Yang Bilibili Inc: Solid Growth But
More Investment (5 Mar 2026)
Kuaishou Technology Kuaishou’s Kling is a leading text-to-video model being ranked first on
Artificial Analysis. Kuaishou has reported ARR reaching US$240mn as of
Dec 2025.
Liu, Yang Kuaishou Technology: Kling
ARR Beat (13 Jan 2026)
Beijing Geekplus
Technology Co., Ltd.
Geekplus is the largest global warehouse fulfillment solution provider. We
believe that AI will empower the company’s algorithm of its warehouse
automation solution, based on its huge accumulated data. In addition, the
company launched its first wheel-based humanoid robot in early 2026,
with the strategy to greatly develop embodied AI in the long term.
Zhong, Sheng China's Emerging Frontiers:
Geekplus: Robots Defining
Warehouses (16 Oct 2025)
M Asia Pacific Insight
16
Futu Holdings Ltd AI shows meaningful benefits to mid and back office efficiency
improvement. Management has indicated efficiency gains were notable
within some departments even though overall headcount looks stable –
mainly driven by the growth in new businesses and markets. Given Futu's
key competitive edge is its R&D, which supports infrastructure and
product development, we see potential for this to be improved by AI's
coding capability. This underpins our assumptions of continued margin
expansion – by about 2ppts a year – if revenue can grow at mid-to-high
teens in the next two years. AI is also becoming a more important part of
Futu’s integrated offerings to clients to empower on information
gathering and analysis. We think this could increase client stickiness and
make client assets more resilient through the cycle.
Huang, Chiyao Futu Holdings Ltd: Multiple
Catalysts Await in 2026 (2 Feb
2026)
Yum China Holdings
Inc.
Yum China embraces AI in all operational functions covering customer
engagement, product development, store management, supply chain,
franchisee management and back office, driving sales growth,
operational efficiency improvements and leads its peers in store growth,
branding, menu innovation and profitability.
Lou, Lillian Yum China Holdings Inc.:
Growth Led by Well-balanced
Strategy; Reiterate OW (5 Feb
2026)
Advantech Edge AI helping accelerate IoT deployment and use cases. Yang, Derrick Advantech: Solid Order Outlook
with Manageable Cost Pressure
(4 Mar 2026)
ZTO Express While digitalization and automation have driven the leading players’ share
gain in China’s express delivery segment, AI will be part of the key
technology tools that leading players leverage to further expand their
competitive advantage in terms of service quality and cost efficiency.
ZTO is the largest player in the segment, with strong balance sheet to
fund innovation, and has been actively combining AI tools with daily
operations and management decision making process. We believe ZTO
will expand its relative advantage vs. peers and continue its market share
gain, powered by the adoption of AI.
Fan, Qianlei ZTO Express: Market Share
Gain Fueling Earnings
Upgrades; Reiterate OW (18
Mar 2026)
Insilico Medicine A leader in AI drug discovery models with a more mature asset-
generation and licensing monetization engine.
Lin, Jack Insilico Medicine: Leading AIDD
Into the Validation Era (2 Feb
2026)
HIKVision Digital
Technology
Launched Guanlan large-scale AI models to provide AI-driven solutions in
security, traffic, and industrial automation.
Meng, Andy Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
China Resources Mixc
Lifestyle Services
Leverage on AI for better understanding on consumption trend and
customer behavior, to customize its marketing and promotion activities to
improve mall retail sales, leading to higher rental income.
Cheung,
Stephen
Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
China TransInfo
Technology
Leverages full-stack AI capabilities for vehicle-road-cloud integration and
smart transportation.
Meng, Andy Greater China Technology
Hardware: Big Smartphone
Downturn Ahead – Prefer Apple
Supply Chain to Android Supply
Chain (22 Mar 2026)
WuXi AppTec AI‑enabled early discovery and data‑driven chemistry accelerate hit
identification and improve R&D productivity.
Tam, Laurence China Healthcare - The WuXi
Group: At the Cutting Edge; Top
Pick WuXi AppTec A (21 Nov
2025)
M Asia Pacific Insight
Morgan Stanl 17
Dian Diagnostics
Group Co Ltd
AI and LLM integration optimize lab workflows, automate test
interpretation and enable scalable patient follow up, creating cost
efficiencies and future data monetization optionality
Yan, Alexis China Healthcare: AI Diffusion
in China Healthcare – The
DeepSeek Effect (10 Mar 2025)
Contemporary
Amperex Technology
Co. Ltd.
Power for AI has emerged as a US$ theme. We see energy storage
as the next wave of growth, with the focus shifting from power availability
to power flexibility. Compelling LCOE and infrastructure deferral value
enhance the economic case for ESS. Data centers could double demand
in five years. We see CATL as a major beneficiary.
Lu, Jack Power for AI: Flexible Power –
The Next Wave of Growth in AI
(17 Mar 2026)
Exhibit 13: Comps table
Share Price Close price date YTD Perf Price Target Upside Mkt Cap Coverage
(Local Curr) (%) (Local Curr) (%) ($ mn) 2026E 2027E 2026E 2027E 2026E 2027E 2026E 2027E 2026E 2027E Analyst
Communication Services
Meitu Inc OW HKD 07/05/2026 (%) HKD 122% 2,778 USD e e e e e e % e % e % e % e Lin, Lydia
Bilibili Inc OW USD 06/05/2026 (%) USD 39% 10,237 USD e e e e e e % e % e % e % e Liu, Yang
Baidu Inc EW USD 06/05/2026 % USD -4% 48,152 USD e e e e e e % e % e NA NA Yu, Gary
Tencent Holdings Ltd. OW HKD 07/05/2026 (%) HKD 36% 565,450 USD e e e e e e % e % e % e % e Yu, Gary
Kuaishou Technology EW HKD 07/05/2026 (%) HKD 14% 27,776 USD e e e e e e % e % e % e % e Liu, Yang
China Literature Ltd EW HKD 07/05/2026 (%) HKD 7% 3,513 USD e e e e e e % e % e % e % e Xu, Rebecca
Mean % %
Median % %
Consumer Discretionary
Alibaba Group Holding OW USD 06/05/2026 (%) USD 27% 335,942 USD e e e e e e % e % e NA NA Yu, Gary
Ecovacs Robotics Co Ltd EW 07/05/2026 (%) 38% 5,627 USD e e e e e e % e % e % e % e Ling, Hildy
Beijing Roborock Technology Co Ltd OW 07/05/2026 (%) 74% 4,481 USD e e e e e e % e % e % e % e Ling, Hildy
Midea Group Co Ltd. OW HKD 07/05/2026 % HKD 21% 88,482 USD e e e e e e % e % e % e % e Lou, Lillian
Haier Smart Home Co Ltd EW HKD 07/05/2026 (%) HKD 15% 28,700 USD e e e e e e % e % e % e % e Lou, Lillian
Yum China Holdings Inc. OW 06/05/2026 % 23% 17,245 e e e e e e % e % e % e % e Lou, Lillian
TAL Education Group OW 06/05/2026 % 40% 2,326 e e () () e % % e NA NA Wang, Eddy
Mean % %
Median % %
Financials
Futu Holdings Ltd OW USD 06/05/2026 % USD 34% 23,754 USD e e e e e e % e % e % e % e Huang, Chiyao
Bank of Ningbo Co. Ltd OW 07/05/2026 % 39% 31,473 USD e e e e e e % e % e % e % e Xu, Richard
Ping An Insurance Group Co of China Ltd OW HKD 07/05/2026 % HKD 39% 156,712 USD e e e e e e % e % e % e % e Xu, Richard
ZhongAn Online P & C Insurance Co Ltd OW HKD 07/05/2026 (%) HKD 49% 2,613 USD e e e e e e % e % e % e % e Zhao, Rick
Mean % %
Median % %
Health Care
Alibaba Health Information Technology UW HKD 07/05/2026 (%) HKD -9% 9,542 USD e e e e e e % e % e % e % e Yan, Alexis
JD Health International Inc. EW HKD 07/05/2026 (%) HKD 42% 19,032 USD e e e e e e % e % e % e % e Yan, Alexis
Dian Diagnostics Group Co Ltd EW 07/05/2026 % -24% 1,881 USD e e e e e e % e % e % e % e Yan, Alexis
Guangzhou Kingmed Diagnostics UW 07/05/2026 (%) -21% 1,889 USD e e e e e e % e % e % e % e Yan, Alexis
Shanghai United Imaging Healthcare Co EW 07/05/2026 (%) 47% 13,462 USD e e e e e e % e % e % e % e Yan, Alexis
Insilico Medicine OW HKD 07/05/2026 % HKD 13% 657 e e e e e e % e % e % e % e Lin, Jack
WuXi AppTec Co Ltd OW HKD 07/05/2026 % HKD 11% 47,777 USD e e e e e e % e % e % e % e Tam, Laurence
Mean % %
Median % %
Industrials
Contemporary Amperex Technology Co. Ltd. OW HKD 07/05/2026 % HKD 23% 317,644 USD e e e e e e % e % e % e % e Lu, Jack
Beijing Geekplus Technology Co., Ltd. OW HKD 07/05/2026 (%) HKD 48% 3,342 USD e e e e e e % e % e NA NA Zhong, Sheng
ZTO Express OW USD 06/05/2026 % USD 10% 20,466 USD e e e e e e % e % e % e % e Fan, Qianlei
Mean % %
Median % %
Information Technology
Beisen Holding Limited OW HKD 07/05/2026 (%) HKD 236% 309 USD e e e e e e % e % e % e % e Lin, Lydia
China TransInfo Technology Co Ltd EW 07/05/2026 (%) 9% 2,127 USD e e e e e e % e % e % e % e Meng, Andy
Advantech OW 07/05/2026 % -10% 12,219 USD e e e e e e % e % e % e % e Yang, Derrick
Dahua Technology Co. Ltd. EW 07/05/2026 (%) 19% 8,827 USD e e e e e e % e % e % e % e Meng, Andy
BYD Electronics OW HKD 07/05/2026 (%) HKD 37% 8,174 USD e e e e e e % e % e % e % e Meng, Andy
Xiaomi Corp OW HKD 07/05/2026 (%) HKD 45% 106,041 USD e e e e e e % e % e % e % e Meng, Andy
HIKVision Digital Technology EW 07/05/2026 % -8% 49,346 USD e e e e e e % e % e % e % e Meng, Andy
Shenzhen Transsion Holdings Co Ltd EW 07/05/2026 (%) 4% 9,639 USD e e e e e e % e % e % e % e Meng, Andy
Foxconn Industrial Internet Co. Ltd. OW 07/05/2026 % 29% 176,967 USD e e e e e e % e % e % e % e Shih, Sharon
Mean % %
Median % %
Real Estate
China Resources Mixc Lifestyle Services OW HKD 07/05/2026 % HKD 12% 14,416 USD e e e e e e % e % e % e % e Cheung, Stephen
Mean % %
Median % %
ROE Div Yield
Company name ` MS
P/E P/B EV/EBITDA
Source: Factset, Morgan Stanley Research
M Asia Pacific Insight
18
What Does AI Adoption Mean for China's Economy?
Robin Xing, Jenny Zheng
In our last China AI Bluepaper, we argued that while AI offers long‑term productivity gains
and could help offset demographic headwinds, near-term support would mainly come
from an AI capex boom, yet labor market disruptions during transition could be a risk to
watch. Rapid AI adoption is now bringing this dynamic into focus. We estimate AI could lift
China’s total factor productivity growth by around 3ppt cumulatively over the next
decade, raising the level of potential GDP by about by 2035 relative to a no‑AI
trajectory. Near‑term net contributions to headline growth, however, would be muted, as
gains from AI‑related capex and early productivity improvements could be largely offset
by labour adjustment frictions. Although China’s aggregate AI job exposure is lower than
in the US, weak demand conditions mean displacement effects could materialise faster
than job creation over the next 2–3 years. Policy mitigation is expected to step up later
this year, but it may only offer a partial offset, as Beijing could exercise greater caution in
labor‑intensive services while allow faster adoption in higher‑end cognitive roles, to
maintain China's competitiveness in GenAI.
AI+ is elevated as the centerpiece in China's 2026-30 Five
Year Plan...
AI is deemed imperative for China’s medium‑term growth. With population aging and a
shrinking labor force, China’s growth model increasingly depends on raising output per
worker. AI is therefore positioned as a macro‑level productivity lever over the longer run
— enhancing labor efficiency, reallocating human capital toward higher value‑added
activities, and accelerating innovation diffusion across sectors. Beyond growth arithmetic,
persistent US–China strategic competition has further elevated AI’s role, reinforcing
Beijing’s focus on technology self‑sufficiency and supply‑chain resilience.
The 15th Five-Year Plan (FYP) institutionalizes “AI+” as a productivity strategy. It
embeds AI across core development objectives, explicitly linking digital and intelligent
upgrading to sustained productivity gains and targeting labor‑productivity growth that
outpaces GDP growth. AI+ is integrated throughout the planning framework— from
computing infrastructure and data governance to large‑scale adoption across industry,
services, and public administration — signaling a clear shift from frontier breakthroughs
toward deployment at scale. At the same time, the Plan strengthens alignment around
domestic AI supply chains and efficiency‑driven development paths, encouraging
hardware–software co‑optimization, open‑source models, and cost‑effective diffusion.
Together with deeper integration into national standards and governance frameworks, this
approach positions AI as a general‑purpose enabler of structural upgrading, supporting
medium‑term growth and extending China’s influence over global AI deployment norms.
M Asia Pacific Insight
Morgan Stanl 19
Exhibit 14: Rising demand for labor productivity boost given
aging population
%
%
%
%
%
%
1996-20 2001-05 2006-10 2011-15 2016-20 2021-25 2026-30E
5Y CAGR
Working Age Population
Employment
Source: Haver, NBS, UN, Morgan Stanley Research. E=UN estimate
Exhibit 15: The 15th FYP elevated "AI+" as a cross-cutting
productivity strategy
Source: Government website, Morgan Stanley Research
...anchoring the AI capex upcycle in the near term
Given that AI is no longer treated as a standalone emerging industry, but as a foundational
input to future growth, akin to digital and physical infrastructure, we expect more AI-
related capex in compute, data centers, energy, and downstream deployment across the
real economy over coming few years. Our Internet team expects China's major technology
companies to increase AI-related capex by 38% this year, to Rmb597bn (US$85bn), and
further up to Rmb711bn (US$101bn) by 2030 (10% CAGR over 2025-30). Our 1H26 China
CIO Survey also shows that the share of AI in IT spending would remain high at % in
2026 (vs. % in 2025), and 40% surveyed CIOs expected to launch their first AI projects
within 12 months. We thus estimate that the strong AI-driven capex growth in emerging
industries to bring a boost to real GDP growth in next two years, offsetting
the drag from continued sluggish property investment.
Exhibit 16: Hyperscalers to step up AI-related capex spending
significantly in next two years
-
100,000
200,000
300,000
400,000
500,000
600,000
700,000
2023 2024 2025 2026 2027 2028 2029
Alibaba Tencent Baidu Bytedance Kuaishou
Source: Company data, Morgan Stanley Research estimates.
Exhibit 17: 1H26 China CIO Survey: AI share in IT spending
budget to remained high at % this year
Source: AlphaWise, Morgan Stanley Research
But productivity gains take time to emerge
Recent international studies (here and here) and historical innovation waves suggest that
M Asia Pacific Insight
20
major technological breakthroughs typically generate a J‑curve in productivity rather than
an immediate uplift. Even if the speed, breadth, and depth of AI diffusion are likely to
exceed those of earlier innovations, the underlying dynamics would be similar. In the near
term, AI adoption entails adjustment costs — ranging from learning frictions and process
disruption to the need for complementary investments in capital, skills, and organizational
change — temporarily weighing on measured productivity. Sustained gains in total factor
productivity typically emerge only later, as firms restructure workflows, labor reallocates
toward higher‑value tasks, and more effective human–AI collaboration are embedded
across a broader set of tasks, functions, and decision-making processes.
Near-term job disruption risk in focus
China has relatively lower job market exposure to AI than the US: According to the
OECD, jobs with high AI intensity are those concentrated in knowledge – and
information‑based services, such as IT, finance, media, and scientific R&D – and jobs with
medium-high AI intensity, such as professional services and education. In contrast, jobs
with low AI intensity are in construction, hospitality, traditional manufacturing, and
primary sectors. Applying this taxonomy, only about 12% of China’s total employment falls
into medium‑high or high AI‑intensity sectors, as compared to ~30% in the US, while
share of employment in jobs with low AI intensity is 38% in China and 15% in the US.
But China's rapid AI diffusion and weak demand starting point suggest that near‑term
disruption risks should not be underestimated. The net labor impact of AI depends on
the interaction between adoption speed, job destruction, re‑skilling, new job creation, and
policy mitigation. Our latest China AI Survey shows that AI adoption has broadened
rapidly over the past year. Against a backdrop of weak corporate earnings, firms in China
may be more inclined to substitute labor with AI to protect margins, potentially leading to
faster job displacement than job creation over the next 2-3 years. This risk is amplified by
the broad‑based nature of AI‑driven substitution: GenAI is displacing junior white‑collar
work (exacerbating the already elevated youth unemployment), agentic AI is beginning to
encroach on mid‑level professional roles (adding pressures to the already stressed
middle-income group), and physical AI – from drones to humanoid robots and
autonomous vehicles – threatens parts of labor‑intensive service jobs, the main absorber
of employment losses in recent years. If not managed properly, these dynamics could
reinforce deflationary pressures, weigh on corporate profitability, and ultimately constrain
resources available for AI innovation.
• Elevated youth unemployment: Global empirical studies (here and here) suggest
that junior‑level white‑collar roles are among the most exposed to AI replacement
risk. In China, youth unemployment has stayed in the high-teens over the past two
years, even after the NBS excluded job‑seeking students from the headline
measure in December 2023. Meanwhile, we estimate the number of university
graduates could rise from in 2025 to by 2030, with a continued
preference for white‑collar employment. If AI adoption outpaces graduate
reskilling and new job creation, pressures on youth employment are likely to
intensify.
• Stressed middle income group: Middle‑income households are already under
strain from slowing wage growth and negative housing wealth effects. Given that
M Asia Pacific Insight
Morgan Stanl 21
knowledge – and information‑based sectors – the typical higher-paid, middle-
income jobs – are more exposed to the rising agentic AI, should near‑term labor
displacement outweigh complementary effects, wage and job pressures could rise,
further weighing on consumption and reinforcing the housing downturn.
• Rising automation risk in lower‑end service jobs. Contrary to perceptions that
these roles are insulated, labor‑intensive services, such as food delivery and
ride‑hailing, are increasingly exposed to automation. Delivery drones and
autonomous driving pilots – rolled out by firms including Meituan, , and
Baidu Apollo Go – highlight that even traditional employment buffers for low‑skill
labor face rising substitution risks, adding to concerns over broader job stability.
Exhibit 18: China's job market is less exposed to AI diffusion
than the US...
0%
10%
20%
30%
40%
50%
60%
Low Low-medium Medium Medium-high High
AI Intensity by Sector
China US
Source: OECD, CEIC, Morgan Stanley Research
Exhibit 19: ...but weaker corporate earnings in China means
higher risk of AI-induced labor replacement
-20% -15% -10% -5% 0% 5% 10% 15% 20% 25%
Wholesale Trade
Professional Services
Agriculture
Health Care and Social Assistance
Retail Trade
Construction
Accommodation and Food Services
Transportation and Warehousing
Mining
Manufacturing
Finance and Insurance
Real Estate
Utilities
Information
Net Profit Margin of Listed Companies by Sector (2025)
China US
Source: Factset, Morgan Stanley Research
Exhibit 20:
Broad‑based nature of AI substitution risk exacerbating deflationary pressures in China
Source: Morgan Stanley Research
What to expect on policy mitigation?
The 15th FYP signals caution on AI‑induced job disruptions: Against concerns over job
displacement, rising income inequality, and social stability, the Plan emphasizes human-
M Asia Pacific Insight
22
machine collaboration, enterprise‑led upskilling, and large‑scale vocational training to
address structural labor mismatches. The directive to promote the application of embodied
intelligence in positions facing labor shortages and high‑risk working environments further
highlights a preference for selective, context‑specific automation where substitution is
most defensible, rather than broad‑based labor displacement.
Policy protection, however, is likely to tilt toward labor‑intensive service sectors. As
Beijing seeks to narrow the gap with global peers in Gen‑AI adoption among higher‑end,
knowledge‑intensive white‑collar roles, it may grant firms greater flexibility in labor
adjustment to accelerate diffusion and productivity gains. By contrast, in areas where
China is already near the technological frontier – such as AI‑enabled delivery drones and
autonomous driving – and where adoption directly affects large pools of service‑sector
employment, policymakers are likely to proceed more cautiously, trying to strike a balance
between technological progress and labor‑market stability.
Expecting more concrete mitigating measures later in the year: Beijing disclosed in late
January that it is contemplating a document addressing the impact of AI on employment.
Consistent with the 15th FYP’s framework, possible measures include:
• Expanded support for job reskilling and vocational training, particularly to
facilitate worker transitions alongside AI adoption.
• Policies fostering new high‑tech and modern service‑sector job creation to
absorb displaced labor and upgrade employment structure.
• Strengthened social protections, including broader unemployment, healthcare
and pension coverage for flexible and gig‑economy workers.
• More calibrated automation in labor‑intensive service sectors, such as delivery,
healthcare support, and consumer services.
Quantifying AI's impact on China's economy
Near‑term effects are likely neutral or even slightly negative, with positives from AI-
related capex cycle and modest initial productivity gains... For 2026-27, the primary
growth impulse from AI will stem from a capex expansion cycle spanning computing
infrastructure, data centers, energy, and downstream deployment. We estimate this could
add to annual real GDP growth. Early productivity gains, however, are likely to
be modest, with task‑level efficiency gains in a narrow set of knowledge‑intensive
activities rather than broad‑based, economy‑wide uplift, as adoption remains uneven and
organizational adjustment incomplete.
…largely offset by transitional labor disruptions: Given the starting point of sluggish
corporate earnings, AI displacement effect will likely well outweigh complementarity
effects and new job creation over next 2-3 years, particularly for junior level knowledge‑
and information‑based jobs. More policy mitigation are expected, but the protection will
likely lean towards lower-end, labor-intensive service jobs, not white-collar jobs. This
means AI's net contribution to headline GDP growth will likely remain largely muted in
2026-27.
Over the longer term, AI would emerge as a structural driver of productivity and
M Asia Pacific Insight
Morgan Stanl 23
potential growth: As AI capabilities mature and diffusion across the economy, we expect
the technology to follow a trajectory consistent with prior transformative technological
shifts: the emergence of new industries, the reshaping of market leadership, and
deepening human-AI collaboration across enterprise workflows. Critically, this dynamic
address one of China's structural headwinds – a shrinking working age population. We
estimate AI could cumulatively lift China's Total Factor Productivity (TFP) by ~3pp
over the next decade, with gains meaningfully back-loaded as adoption scales across the
economy. Combined with AI-related capex upcycle, we expect a potential GDP level
~ higher by 2035 compared to a trajectory absent AI adoption.
Exhibit 21: AI's impact on China's GDP growth: Near-term neutral, long-term positive
%
%
%
%
%
%
%
%
%
%
2026-27E 2028-30E 2031-35E
AI Boost to China's Real GDP Growth, ppt
Capital Input Labor Quantity Labor Quality TFP Overall Growth
Source: CEIC, Morgan Stanley Research estimates
Risks to watch
Key swing factors to our baseline forecast are AI diffusion speed & scale and policy
reaction
• Downside risk – Disruptive AI replacement across the economy, limited policy
mitigation: Under this scenario, AI diffusion accelerated across both junior
white‑collar roles and labor‑intensive services, such as autonomous driving,
delivery drones, and service robots, while policy mitigation is limited, with focus
more skewed towards tech upgrading over labor market stabilization and social
welfare supports. Under the prevailing weak demand environment, this means firm
would tend to deploy AI primarily as a cost‑cutting tool, translating productivity
gains into massive layoffs than output gains. This would add to deflationary
pressures, eroding corporate earnings and possibly constraining firms ability to
invest beyond essential AI adoption. In turn, a negative feedback loop would be
formed, in which AI boosts micro efficiency but acts as a macro demand headwind,
delaying the emergence of longer‑term productivity gains.
• Upside risk – More measured AI diffusion, more demand-centric policy
direction: Under this scenario, AI adoption remains rapid but more gradual and
augmentation‑focused, with firms prioritizing productivity enhancement over
outright labor replacement. At the same time, policy response is more proactive,
M Asia Pacific Insight
24
combining reskilling, vocational training, and stronger social protection with a
clearer shift toward consumption‑led rebalancing. Improved income security limits
precautionary saving, supports consumption, and helps stabilize the housing
market, easing deflationary pressures. Stronger demand allows productivity gains
to flow through to corporate earnings rather than pure cost compression, enabling
firms to scale AI capex alongside complementary investments in organization, data,
and human‑AI workflows. As profitability improves, new tasks and roles emerge
more quickly, allowing job creation to catch up or even outpace displacement. In
turn, AI becomes a reinforcing force for growth – supporting a virtuous cycle of
firmer demand, stronger earnings, higher investment, and faster innovation
diffusion.
M Asia Pacific Insight
Morgan Stanl 25
Key Developments in Strength – Power, Humanoids,
Autonomous Driving
Power
Eva Hou, Jack Lu, Tom Li
Flexible Power – The Next Wave of Growth in AI
China Power and Grid Equipment: Global Supercycle Ongoing; Looking for the Next Leg
Up
Power for AI has emerged as a US$ theme. We see energy storage as the next wave of
growth, with the focus shifting from power availability to power flexibility. Compelling
levelised cost of energy (LCOE) and infrastructure deferral value enhance the economic
case for ESS. Demand from data centers could double in five years.
The next wave of change around "Powering AI": The quest to power the AI boom has
significantly re-rated the global power value chain, creating a market worth over US$
in the past two years. This re-rating has catalyzed a rotation from nuclear power to grid
facilities, backup/gas generators, and more recently to fuel cells. While power capacity is
facing supply chain constraints across the board, we believe the next opportunity will be
in energy storage systems (ESS), as investment in AI infrastructure continues to accelerate
and inference workloads inflect, shifting the focus on AI from energy availability to power
volatility, system flexibility, and time-to-power. We believe the thematic rotation toward
energy storage is not just a logical step but a necessity. It is not about energy storage
replacing generators, but the system rebalancing itself from one that built ahead of
demand to one that creates a buffer against future demand, using inventory.
Exhibit 22: Global Power for AI Basket: >US$1tr market value
accretion
Source: Factset, Morgan Stanley Research
Exhibit 23: Segment rotations from nuclear, gas/backup
generators to fuel cells; the next opportunity should be energy
storage
Source: Factset, Morgan Stanley Research
M Asia Pacific Insight
26
Exhibit 24:
Power consumption shifting from training workload to inference
Source: Morgan Stanley Research
ESS cost advantages beyond LCOE comparisons: The ESS cost advantage can be seen
not only in its LCOE, but more importantly in the ability of electricity inventory to defer or
downsize the high opportunity costs of traditional capital-intensive power infrastructure.
This time-to-power flexibility and infrastructure deferral value meaningfully enhance the
economic case for storage. Battery costs will likely continue to evolve with the coming of
the sodium-ion era – cheaper and safer, with the ability to deliver steady energy at low
temperatures.
M Asia Pacific Insight
Morgan Stanl 27
Exhibit 25: Solar integrated ESS LCOE across regions vs. gas power LCOE across regions
Source: Morgan Stanley Research estimates
Exhibit 26: Battery cost evolution driving "inventory" building for power
Source: Wind, Morgan Stanley Research estimates
Quantification: As AI data centers develop at an unprecedented pace and scale, the
binding constraint is no longer demand or capital, but time‑to‑power and peak‑load
management, in our view. ESS will likely emerge as a natural, economically efficient
approach at a turning point in the cycle. We map out global ESS demand from data
centers in the next five years, considering: 1) peak-shaving demand from inference, and 2)
deployment required to mitigate shortages in traditional power capacity. We forecast
global ESS annual incremental deployment from data centers at ~321GWh by 2030 –
169GWh in the US, 85GWh in China, and 68GWh in ROW, vs. a global power market utility
scale of 325GWh in 2025, which we also expect to increase at a 22% CAGR by 2030e,
driven by the economic energy transition. Overall, we expect a 30% CAGR in global ESS
M Asia Pacific Insight
28
annual incremental deployment by 2030e.
Exhibit 27: Global ESS annual incremental deployment forecast
for data centers...
Source: Rystad, Morgan Stanley Research (e) estimates
Exhibit 28: ...including utility-scale ESS in power markets
Source: Rystad, Morgan Stanley Research (e) estimates
Exhibit 29:
Heat map of potential beneficiaries in power-for-AI ESS
Source: Morgan Stanley Research
M Asia Pacific Insight
Morgan Stanl 29
Exhibit 30: Bull/base/bear case for ESS shipment forecasts
Source: Morgan Stanley Research estimate
Besides ESS, we are bullish on gas turbines and high-voltage grid equipment (such as
transformers). With AIDC boom, the global electricity system is shifting from stagnation
to expansion, amid inadequate infrastructure, leading to a multi-year export upcycle for
China. We believe global gas turbine demand is likely to increase from ~60GW . to
>100GW in the next ten years, with higher ASP. In addition, on the grid side, we foresee
grid capex CAGRs of 9%/12%/% for China/Europe/US in 2025-27E, respectively,
compared with the global grid capex CAGR of 6-7% in 2020-25, driving the demand for
large power transformers.
On the supply side, we view equipment makers' capex plans as still disciplined, as the
capex-to-sales ratio remains largely stable. This should continue to foster a sustainable
uptrend in lead time (2-3x vs. the normal level) and pricing (up 30-50% in 2025).
Therefore, we see sufficient upside surprise for China's equipment exporters along the gas
turbine and transformer supply chains. Our most preferred plays are Yingliu (gas turbine
vanes and blades) and Sieyuan (power transformer and HV switchgear). We expect a 40-
70% earnings CAGR in 2025-27E for these two companies, underpinned by solid backlog
and rising global market share. Our analysis indicates that every 1ppt global share gain can
lead to 21-31% EPS upside in 2027E for both companies. On valuation, they are trading at
close to or below PEG in 2026E, which seems attractive compared to global peers at
~ on average.
M Asia Pacific Insight
30
Exhibit 31: ASP of GE Vernova's gas turbine orders
Source: Company data, Morgan Stanley Research. Note: The pricing also includes other non-Gas Power
businesses (steam, nuclear, misc.).
Exhibit 32: Grid capex in North America, Europe, and China in
2020-27E
Source: IEA, S&P Global Market Intelligence, CEC, Morgan Stanley Research estimates
Humanoid / Robotics
Sheng Zhong
Focus shifting from locomotion to "brain". China humanoids' locomotion capability
improved meaningfully in 2025, becoming the de facto leader, highlighted by the
impressive performance at the 2026 Spring Festival Gala,
