Description

Our bear case is a 22% IRR, yet this case does not give management credit for their 2yr plan of mid-20% topline and ~30% EBITDA growth that was laid out ~15 months ago (May 2019). This plan was laid out prior to Blackstone’s $150m pref investment (announced June 22; 4.5% coupon / $17 conversion price). Even more importantly, as we are seeing across our sectors, Covid has massively accelerated the pace of enterprise adoption of cloud architecture/investment in digital infrastructure and significant budget expansion of the major cloud providers dedicated to cloud infrastructure buildouts (BABA $28bn over the next 3yrs and Tencent $72bn over the next 5yrs). The company has confirmed incremental demand from “scaled retail” such as financial institutions, online education, online gaming, and streaming customers (Bytedance, Meituan, PDD, etc.) over the last 5mths. There is a very high likelihood that our Base case is significantly too conservative.

Blackstone's level of operational and financial diligence is important because VNET is a Chinese SMID-cap that has a history of short seller accusations (2014, largely refuted) and execution missteps. New management turned over in early 2018 after selling their legacy CDN/connectivity business (for $1), which was a major distraction and headwind to growth/profitability, to focus on the IDCs (“Internet Data Centers”). Based on our conversations with the sell side in China, dating back to the short seller report, large global and Chinese/HK funds have been reluctant to invest in the company. The Blackstone stamp of approval will change this. GDS trades at 30x Fwd EBITDA and 23x 2yr Fwd. As seen above, our multiples are safely conservative (probably to a fault) and will still provide very attractive IRRs. 

Thesis: (A) Sustained secular growth of digital infrastructure in China

1.    China is 3-5yrs behind the US in enterprise cloud migration and is set to accelerate post-covid. Software spend in China is only ~6% of current US-levels and the overwhelming majority of software applications are still on-premise.

2.    Greater at-home entertainment and collaboration require greater digital infra spend after seeing new highs in the peak demand during Covid.

3.    Rollout of 5G, adoption of IoT, automation, increased data consumption, and demands for low latency means strong IDC demand in tier 1 cities will persist.

4.    On top of govt plan of $3.8trn investment in digital infra over the next 5yrs ($63bn/mth), BABA announced $28bn over the next 3yrs and Tencent $70bn over the next 5yrs directly into cloud infrastructure expansion. Chinese data centers are a primary beneficiary of this.

(B) Sell Side models are wildly conservative on margins

The Management Plan is for rev growth of 23%/25% in 2020/21 and EBITDA growth of 24%/37.5% in 2020/21. The Sell side models are currently giving VNET credit for that revenue growth but only 20%/34% growth for EBITDA. One prominent sell side model assumes margin will stay constant at 2019 levels for the next 5yrs (screenshot below). Based on our conversations with sell side analysts, this is due to embedded conservatism as management execution is early and they want to set the bar appropriately. Incremental margins for data centers are typically as high as 60-70% given the significant fixed cost leverage base in turning on incremental racks in an already-built DC. We are confident that the company will see margin expansion and are modeling 33% EBITDA margins in 2022.

Sell Side Model:

(C) Ability to invest at attractive ROICs supported by new govt financing policies 

VNET has been able to achieve 15-20% ROICs / 40-50% EBITDA margins in their retail data centers at maturity (55-80% utilization and takes 18-24mths depending on location). Wholesale (cloud providers like BABA/Tencent) customers require higher upfront capex b/c of the higher power density racks, but higher steady-state margins for less hands-on customers and 10-15% ROICs. Mature US data center players have maintained LDD ROICs, with EQIX closer to mid-teens (due to more retail and interconnection exposure). China is 3-5yrs behind the US in adoption, so these modeled VNET ROICs are reasonably conservative for the next several years.

Mgmt confirmed that the Govt's New Infra Policy is seeing support from domestic banks with more favorable terms on project finance. In the past, the IDC was viewed like C&I - if there no land and building on site, banks wouldn’t provide project finance so VNET has had to resort to more expensive capital when doing development. Historically, they had to issue US Bonds at ~7% rate; now they are receiving quotes at 5-6% rates in local currency with 5-7yr terms and ~2/3 leverage. This directly reduces WACC and decreases equity drags on construction.

As for the sustainably of ROICs - mgmt walked us through their BABA relationship - BABA/Tencent look at the fundamentals of the DC companies and get comfort that they will survive for the next 8-10yrs and have the resources+capital to grow with them. Building out space in an IDC is a long-term investment and BABA wants a trusted partner and is not looking to push price to the lowest common denominator; they are happy to have the infra provider grow and maintain low-teens ROICs while their capital-light business sustains multiples of that. Based on our expert calls there are really only 4-5 scaled/trusted independent IDCs in China - GDS (tier 1); VNET, Beijing Sinnet, AtHub (tier 2); Dr. Peng (tier 3) with commentary unanimously saying VNET is gaining popularity and Sinnet losing share.  

(D) Blackstone Stamp of Approval for a Chinese SMID-cap trading at a 60% discount to GDS/US DC peers.

Discussed above, but Blackstone now has a non-voting board observer seat and very strong relationships in China, validating the company’s pipeline and credibiilty of the mgmt team.

(E) Secondary Listing in HK

Feedback from contacts in China has been hesitancy from Asian funds given rhetoric around delisting Chinese companies from the US exchanges. There have been rumors about a secondary listing in HK and the company appears to have hired bankers and lawyers to look into the prospect. We have seen several other companies’ stock prices rewarded for doing so.

We ran a DCF to sanity check our near-term valuations given the magnitude of the importance of terminal value in data center investments; we get to a $60 stock price.

VNET currently trades at a 13% yield on our 2023 'Steady-State' Unlevered FCF:

2024E is for 25% 'Steady-State' FCF growth. That should translate into at least a 25x FCF multiple, which implies ~$70 stock price in a couple years:

Appendix: Comparing GDS 2017A to Base Case 2022E for VNET: Embedded Conservatism

1.    VNET ended 2019 with 10% less cabinets than GDS in 2017. We are forecasting VNET’s cabinets to grow 30k over the next 3yrs versus GDS, which grew cabinets by 38k in 2yrs.

2.    GDS ended 2017 w/ 47% Gross Margin. We are only forecasting VNET to have 45% GMs in 2022

a.    VNET actually takes in 2x the gross profit dollars per customer given it’s more heavy retail exposure that require more value-added services (at lower margins)

3.    Our 2022E EBITDA margin for VNET is in line with where GDS was in 2017 and 1100bps below GDS in 2019

4.    Slightly more conservative capex assumptions relative to GDS in 2019A.

Catalyst

1. Earnings execution and upward revisions of sell side numbers

2. Secondary HK Listing

3. Chinese govt announcements for new data center financing support and relaxed power/carbon quotas

4. Announcements of large wholesale contract wins