+ FY20 NPI and DPU grew 37.7% and 20.5% YoY respectively, led by newly-acquired KDC4 and DC1 in Singapore (in 2H19) and Kelsterbach DC in Germany (on 1 May 2020). Portfolio occupancy was also higher from completed AEI and higher demand.
+ Portfolio occupancy improved QoQ from 96.7% to 97.8% (FY19: 94.9%). Higher occupancy in 4Q20 stemmed from: 1) the handover of newly converted DC space at KDC5 which lifted occupancy from 84.2% to 100%; 2) tenant expansion at KDC1 which increased occupancy from 89.2% to 91.1%; and 3) a new tenant at KDC2 which raised occupancy from 93.5% to 98.2%. AEI to bring more power onsite at Dub1 also allowed KDC to lease out additional space, bumping up its occupancy from 61.8% to 81.1%.
+ Acquired Amsterdam DC on 24 Dec 2020 for S$48.1mn; initial NPI yield of 5.1%. Amsterdam DC comprises a shell & core data centre and an office component, with occupancy of 99.1%. The asset is located near the Amsterdam Internet Exchange, one of the world’s largest hubs in terms of connections and traffic. Amsterdam DC has an occupancy of 99.1% and is leased to data-centre and IT service firms.
– Natural decay of leases reduced WALE to 6.8 years (FY19: 8.6 years). Colocation (Colo) leases are the most profitable leases as rents include charges for facility management and the rental of M&E equipment. However, Colo leases typically run on shorter WALEs of 2.7 years vs. 11.2 years for fully-fitted and 7.3 years for shell & core data centres. Fortuitously, most of KDC’s Colo leases are in Singapore, which is a rising rental market due to limited new supply. As such, shorter WALEs should allow KDC to capture higher rents as the leases are marked to market upon renewal. In contrast, with a WALE of 1.5% and persistently low occupancy of 63.1%, KDC’s struggling Malaysian asset, Basis Bay (0.8% of AUM), may face occupancy risks should its existing tenant decide not to renew.
Evaluating potential acquisitions. KDC is evaluating several piecemeal and portfolio acquisitions with cap rates of 5-7%. However, travel restrictions in several countries have prevented KDC from physically inspecting the assets overseas. Management has found a way around this, which is to focus on shell & core assets which are more passive in nature and rely on third-party valuers. Using this strategy, KDC was able to acquire Amsterdam DC before the year ended.
Demand-supply gap to push up market rents. Singapore is KDC’s core market, accounting for 56% of its AUM. Given the moratorium on data centres in Singapore, we expect market rents to be bid up in the coming two years. Present higher demand has led to higher occupancy for KDC’s portfolio, though market rents have not moved. While KDC’s portfolio occupancy in Singapore is high at 97.0%, Colo leases in Singapore have WALEs of 1.4-4.0 years. These coincide with expected rent appreciation. In the meantime, KDC is expected to benefit from organic growth as many of its leases have built-in periodic rental escalations averaging 2-4% p.a.
Upgrade to ACCUMULATE, DDM-based TP raised from S$2.91 to S$3.20
Changes from our previous report
Our DDM-based TP has been raised to reflect higher occupancy and asset productivity following AEI, which raises our NPI by 6.9% on a same store basis (excluding any acquisition assumption). Our previous TP of S$2.91 assumes that KDC will make a S$500mn of acquisitions in 1Q21e (NPI yield 6% and LTV 30%). In this report, we push back our S$500mn acquisition assumption (NPI yield 6% and LTV 30%) from 1Q21 to 4Q21.
Scenario analysis: with and without acquisitions
Our acquisition assumption reduces FY21e DPU by 5.4% due to an enlarged share base, but increases FY22e DPU by 5.3% (Figure 1).
Demand for data centres is supported by increasing 5G, smartphone and cloud adoption. KDC’s higher P/NAV of 2.4x (Figure 3) is supported by the forecast growth in data centres and its ability to deliver earnings growth through AEI, rental improvements and accretive acquisitions. We forecast DPU yields of 3.3%/3.8% for FY20e/21e, which should deliver a forward yield spread of 220bps, the average of its five-year historical yield spread over 10YSGS (Figure 2).