1. Investment thesis: Acquire China's largest offline commercial operating network
Xindameng's most valuable assets can be abstracted into four layers:
529 commercial nodes × about 40,000 employees × hundreds of thousands of merchant relationships × daily ongoing operating activities.
Wanda currently excels at “building plazas, attracting tenants, opening, and operating.”
Goal after FDEPE takes over:
Transform 529 Wanda plazas into a commercial operating machine that can continuously learn, continuously optimize, and continuously replicate.
PE handles capital, governance, and incentives; FDE goes directly to the front line, embedding operating experience into the system.
2. Establish “Newland FDE”: report directly to the CEO
The headquarters will set up a core FDE team of about 50 people:
AI engineers + data engineers + product managers + commercial operation experts + leasing experts + finance staff.
The first batch will go directly into 30 Wanda plazas, working together with general managers, leasing, operations, property, marketing, and finance.
Each FDE answers only one question:
How much extra profit did the mall I’m responsible for generate this month?
System development is fully tied to operating metrics.
3. Build the “Wanda Operating Graph”
Unify all malls into a single operational data structure:
Mall → Floor → Stall → Brand → Store → Contract → Rent → Sales → Foot traffic → Membership → Promotion → Parking → Energy consumption → Staff.
Also integrate weather, holidays, city events, shopping district demographics, competing projects, and online consumption trends.
Goal:
Each stall has its own dynamic operating account.
Management can continuously assess:
- Which stall is creating value?
- Which brand's rent is mismatched with its operational capability?
- Which area’s sales per square meter is continuously declining?
- Which tenant faces exit risk in the next 90 days?
- Which vacant stall should be filled with which brand?
IV. Deploy six directly revenue-generating agents
- Leasing Agent | Task: brand discovery, matching, leasing | Core metric: vacancy rate
- Rent Agent | Task: dynamically calculate store space value | Core metric: rent per sqm
- Merchant Growth Agent | Task: help merchants improve operations | Core metric: merchant GMV
- Marketing Agent | Task: events, membership, advertising | Core metric: ROI
- Energy Agent | Task: air conditioning, lighting, equipment optimization | Core metric: energy consumption per sqm
- Operations Agent | Task: daily diagnosis of mall operations | Core metric: NOI
Among them, the most important is the Merchant Growth Agent.
Wanda can further participate in merchant growth and explore:
Base management fee + GMV share + advertising revenue + AI operations service fee + incremental profit share.
The logic becomes:
Merchants earn more, and Wanda earns more.
5. Capture an “AI Native Wanda Mall” in 100 days
First select 30 pilot sites:
- First‑tier cities: 5 sites
- Second‑tier cities: 10 sites
- Third‑ and fourth‑tier cities: 10 sites
- Inefficient / projects pending renovation: 5 sites
Suggested 100‑day challenge target:
- NOI increase of 5%–8%
- Vacancy period reduction of 30%
- Energy cost reduction of 8%–12%
- Marketing expense efficiency improvement of 20%
- Manual reporting workload reduction of 50%
- Average number of brands managed per leasing staff increased by 30%
These figures are only for experimental targets and are not recorded as investment returns in advance.
Each validated business action results in:
Wanda Playbook 001 / 002 / 003……
Then 30 stores → 150 stores → all plazas.
6. Turn operating experience into corporate assets
Wanda has accumulated a large amount of tacit knowledge over more than twenty years:
- Which cities are suitable for which formats;
- Which locations are suitable for which brands;
- How much rent a certain brand should pay;
- Which brand combinations can drive traffic to each other;
- What should be completed 180 days before opening.
FDE captures each:
Judgment → Action → Result
Structured recording.
Ultimately yields:
Wanda Commercial Model
Input:
City + Business district + Building + Population + Competition + Consumption
Output:
Business format mix + Brand mix + Rent + Leasing order + Marketing budget + Staffing + Projected NOI.
New projects first run the model, then organize people to execute it.
VII. Second growth curve: Commercial operating infrastructure
When all plazas form a unified operating network, four types of businesses can emerge:
Wanda Ads
National commercial space advertising network.
Wanda Merchant Cloud
Providing operating agents to merchants.
Wanda Intelligence
Export leasing, marketing, pricing, and property capabilities to other shopping centers.
Wanda Expansion
Use models to identify third‑party commercial projects suitable for acquisition.
Create a flywheel:
More malls → more data → better operating models → higher NOI → stronger acquisition capability → more malls.
VIII. Governance mechanisms restructured according to FDEPE
Board focus areas:
- NOI
- Free cash flow
- Revenue per square meter
- Third‑party managed area
- Verifiable incremental profit generated by FDE
Founded:
FDE Value Creation Committee
Answer each month:
How much engineering resources are invested → What operational actions are changed → Ultimately how much cash flow is increased?
Incentives for management and the FDE team are gradually tied to verifiable incremental cash flow.
IX. Three‑year roadmap
0–6 months: proof
30 pilot sites.
There is only one goal:
Demonstrate that FDE can consistently change NOI.
6–18 months: replication
Expand to 150–200 seats, forming a standardized Deployment Kit.
18–36 months: scaling
Cover the existing commercial network and apply the capability to new third‑party projects.
10. The greatest transformation risks and objections (counter‑evidence)
FDEPE must first demonstrate that it will not “ruin” a company that is already profitable.
Objection 1: Wanda is already one of China’s strongest commercial operators; why should 50 FDEs know business better than tens of thousands of veteran employees?
This is the strongest objection.
Attraction, foot traffic, store locations, brand mix, and city consumption contain a large amount of tacit knowledge that cannot be directly digitized.
An excellent leasing manager may need only a glance at a location to know whether a brand can survive.
The model may require several million data points to approximate such judgments.
Therefore, FDE should not start with the premise of “replacing the veteran.”
The correct organizational relationship should be:
The top Wanda operators × FDE.
First identify the company's internal top 1% operators and amplify their judgment ability across the entire organization.
Objection two: 529 plazas are not 529 identical nodes.
Wanda locations in Beijing CBD, Chengdu, Kunming and a county-level city differ greatly in consumer purchasing power, rent structure, brand offerings, customer base, and competitive environment.
Therefore:
Success in one plaza ≠ replication across 529 plazas.
A hierarchical model must be established:
City tier × business district type × plaza lifecycle × customer segment × format structure.
Ultimately, it may result in 20–50 operating prototypes rather than a single nationwide algorithm.
Objection three: Data correlation is easily mistaken for operational causality.
For example:
The system found that “women’s apparel sales are higher near bubble tea brands.”
This does not mean that adding a bubble tea shop can boost women's clothing sales.
Commercial real estate has many common variables:
Location, floor level, foot traffic, city consumer purchasing power, holidays, brand momentum.
Therefore, all major business models must go through:
Forecast → Small-scale experiment → Control group → Financial validation → Scale deployment.
FDEPE must be an operational experiment system, not turn into a large BI project.
Objection four: Optimizing rent may harm the entire ecosystem
If a rent agent discovers that a tenant has strong payment ability and continuously raises rent, short-term NOI will increase.
But after the tenant's profit declines:
Brand exit → Increased vacancy → Decline in consumer experience → Decrease in foot traffic.
Thus the algorithm's objective cannot focus solely on rent optimization.
A more reasonable objective function is:
Property's long‑term NOI × tenant survival rate × consumer value.
It may even require offering low rent to certain strategic brands proactively.
Objection five: tenants may be unwilling to provide real data to Wanda.
If Wanda simultaneously controls:
sales, profit, foot traffic, inventory, membership, marketing effectiveness
tenants will worry that this data will ultimately become a basis for rent increases.
This could directly undermine Merchant Cloud.
Therefore a clear data contract must be established:
Data used to help tenants grow should not become a unilateral basis for raising rent.
Data governance itself is a product.
Objection six: the FDE team could become a new headquarters bureaucracy.
This is one of the biggest organizational risks for FDEPE.
50 individuals could swell into after one year:
Data Center + AI Center + Digitalization Committee + numerous reporting mechanisms.
Ultimately creates many dashboards, but does not generate cash flow.
Therefore, a strict rule must be set:
FDE projects without operating results are automatically terminated.
Each project must have:
Baseline → Action → Control → Result → P&L Attribution。
Objection seven: It is difficult to attribute profits generated by AI
This year, NOI rose 5%, possibly due to:
Economic recovery, lease changes, competitor mall closures, weather, marketing activities, or AI.
If attribution is impossible, the effectiveness of FDEPE cannot be proven.
Thus, it must be established from day one:
FDE Ledger
Each renovation record:
Investment cost / Control group / Operational actions / Incremental revenue / Cost savings / Duration / Confidence level.
Final outcome:
AI investment of 1 yuan, audited to generate how much free cash flow.
This is one of the most important financial infrastructures of FDEPE.
11. The failure modes that truly need vigilance
The most dangerous situation is not that the model is unusable.
But rather three years later:
- Investment of several hundred million yuan
- Build a massive data platform
- Deploy dozens of agents
- Create a beautiful dashboard
- Organize extensive training
- Everyone is using AI
Then discover:
NOI has hardly changed in attributable terms.
Therefore the entire project must adopt PE-style discipline:
30 experiments
If after 6 months it cannot demonstrate clear unit economic improvement:
Stop expansion.
If proven:
Continue investment.
If some scenarios are effective and others are not:
Only expand the effective portion.
The core discipline of FDEPE should be:
Technology has no inherent value. Only verifiable cash flow changes have value.
XII. Wanda Investment Logic in the FDEPE version
These transactions have a chassis that is highly suitable for FDEPE:
Massive offline network + mature operating system + repeatable business nodes + stable cash flow + large amount of tacit knowledge.
Traditional PE value creation mainly relies on:
Capital structure + governance + incentives + cost + mergers and acquisitions + exit.
FDEPE adds:
Engineering capability + AI + operational experiments + knowledge assetization + large‑scale replication.
However, the entire investment logic must not be based on:
"AI can improve Wanda's efficiency."
It must be based on a stricter question:
Can FDE demonstrate a business improvement repeatedly across 30 different types of plazas, and ensure that each yuan of technology investment generates an auditable incremental free cash flow?
If this can be proven, 529 commercial nodes would create a very strong scale leverage.
A single action adds only 1,000,000 yuan of verifiable profit per plaza each year; scaling it to 500 plazas yields roughly 500,000,000 yuan in profit.
Ten such actions represent a business value at the tens of billions of yuan level.
Therefore, what FDEPE is looking for is not a “super AI project.”
What it seeks is:
100 small operating algorithms that can increase cash flow by 100,000, 500,000, and 1,000,000 yuan each year, and then replicate them hundreds of times.
This may be the most worthwhile path for applying FDEPE to Wanda:
PE purchases the cash flow, FDE looks for minor improvements within the cash flow, and then leverages the massive operating network to turn those small improvements into huge value.
I believe this version is closer to the FDEPE investment committee’s view than simply emphasizing “AI transformation of Wanda”: first assume the transformation may fail, then design a mechanism that makes mistakes cheap and success replicable.