Thread by @nsheth12

$GDS reported earnings last night. It’s the leading carrier-neutral data center operator in China, and it’s a company I follow pretty closely. My takeaways from earnings, both the good and bad 👇

1/ First, a bit about the company: GDS operates in T1 cities in China (mainly, Beijing, Shanghai, Shenzhen, Guangzhou) where land & power for DCs within 100km of city center is very scarce.

2/ GDS customers (like BABA, Tencent, etc.) really want DCs near cities, since latency w/ end users is better.

3/ This means that sales & land held for future development are KPIs I watch closely. Sales are important b/c they signal strong customer demand for GDS’ data centers, and growth in land pipeline means that GDS has the scarce land + power resources to fulfill that demand.

4/ The good: GDS beat on both. ✅

In 2020, they had organic sales of 109k sqm (vs. 80k target at beginning of year) and 28k sqm of sales thru M&A (vs. 20k target). This signals strong demand.

5/ Area held for future development was 480k sqm, up from 245k sqm last year. Resources are scarce, so it’s good that GDS is sourcing this land and power ahead of time. At a run rate of 100k sqm organic/year, this is almost 5 years’ worth of land pipeline!

6/ Obviously, we’d go plot by plot of land to really assess the quality of this pipeline (based on location, power allocation, etc.), but I’ll spare you the details here :)

7/ M&A is becoming a bigger part of the strategy & thought they made some nice moves. SZ8 acq. is a great downtown location & price seems good too - but not big enough to move the needle. The acq. in Tianjin is also interesting as the next edge of town location for Beijing.

8/ They’re really starting to talk up SE Asia expansion. The strategy is to track expansion of existing Chinese hyperscale customers into SE Asia. I’m optimistic about this growth vector, esp since Singaporean sovereign wealth owns 1/3rd of the company.

9/ Finally, they’re guiding to 90-100k sqm organic net sales for this year — I think it’ll probably be 110k+.

I also think M&A will probably be much bigger than it was in 2020 and FY21 capex will end up higher than the guide of RMB 12bn.

These guys tend to beat expectations...

10/ This is what Dan Newman said about guidance:

“Look, we haven't factored in regionalization. We haven't factored in M&A beyond the flow deals as I just described them. We haven't factored in a higher level of organic growth...

11/ We haven't factored in potentially bringing forward a CapEx if we see opportunities to acquire land or buildings, which are kind of going to be opportunities that are kind of onetime opportunities that you have to grab when they're there, otherwise, they've gone forever.”

12/ So that was the good. On to the bad...

13/ Given the $ they raised last year, GDS is under pressure to deploy capital & grow faster. Growth has been good, but they may have sacrificed returns on new projects to achieve growth.

14/ Their slides show that cash-on-cash returns are stable, but that doesn’t include this year’s sales, which won’t show up in P&L until future years.

15/ Since GDS is wholesale, their IRRs have always been worse than retail, colo peers like $EQIX. Doesn't seem like that will change soon.

(GDS on left, EQIX on right. 13% cash on cash for GDS, 27% for EQIX)

16/ Other odd thing was GDS switching some build-operator-transfer (BOT) projects back onto their own balance sheet, and retroactively updating metrics like area pre-committed to account for that.

This added something like 14k sqm to their organic net adds for FY20.

17/ BOT projects are typically not in tier 1 zones and are built on customers’ own sites specifically for them. Essentially, GDS is an outsourced DC developer + operator for a specific customer on those projects.

18/ As you can imagine, IRRs on those projects are lower since there is no land + power scarcity for which GDS can charge a premium.

19/ For this reason, GDS historically did those projects to expand customer relationships and then moved them off balance sheet, usually to a 3rd party investor like GIC, in order to avoid tying up capital on lower return projects.

20/ This Q, though, they did a 180 and decided they’d hold the assets for two of the projects (LF10 & HL1).

21/ These are clearly less high quality assets.

Dan mentioned that they need to put 80% leverage on those projects to get an acceptable return on equity. For context, GDS projects are usually only 60% debt financed.

22/ I don’t see a good explanation for this other than that GDS is sacrificing returns in order to show better organic sales nums. Dan was asked about it on earnings call and didn't have a great answer IMO.

Would love to hear if anyone has a better perspective on this.

23/ That aside, it was a fine quarter all around.

I still think GDS is by far the best data center operator in China.

The key things I’ll continue to track are sales, land pipeline, and IRRs.