March 01, 2019 - 5:37pm EST by
2019 2020
Price: 35.78 EPS 0 0
Shares Out. (in M): 287 P/E 0 0
Market Cap (in $M): 10,246 P/FCF 0 0
Net Debt (in $M): 7,977 EBIT 0 0
TEV (in $M): 18,223 TEV/EBIT 0 0
Borrow Cost: General Collateral

Sign up for free guest access to view investment idea with a 45 days delay.





IRM is the leader in physical document storage and retention, sporting low churn rates that provide predictable cash flows.  Think boxes on shelves in a warehouse, or originally in depleted mines hence the name. The Company uses that cash flow to invest in emerging markets and to pivot the company towards high growth datacenters to reinvigorate the top line and EBITDA.  The 2020 goal is to grow EBITDA 5% organically. It sports a 7% dividend yield and trades below 11x 2020E AFFO/share, cheap compared to other REITs.

That’s the bull story.  

Here’s what we believe and why we’re short: a) volume declines seem to be accelerating; b) pivot to datacenters is ill conceived as that market is bifurcating between high quality retail centric datacenters (think EQIX) and build-to-suit for hyperscale datacenters (think CONE) whereas IRM is neither; c) dividend coverage eats up 88% of FCF, forcing IRM to issue debt and equity to fund datacenter requirements; d) poor FCF conversion; and e) EBITDA is the more appropriate way to value this business vs company defined AFFO.


a) Volume declines seem to be accelerating

Organic growth for the whole company has declined to less than 1% in each of the last four quarters, with the prior two slightly contracting, led by a slowdown in sales to new and existing customers while destructions and terminations have stayed steady.  

The trend is most prominent in NA where organic volume has been negative since 2Q17.  Management had decided to emphase price over volume. What’s interesting is that the destructions have started to accelerate, which usually means future volume from existing storage customers should start to decline on a lag.  With destructions accelerating (perhaps because of the storage price increases?) the Company may have to rethink its strategy.


b) Pivot to datacenters ill conceived

We are long datacenters and believe in the multi-year trend of enterprises moving compute and storage to the cloud via public and hybrid public/private datacenters.  Speaking with management teams and industry brokers, the market seems to be bifurcating. On the one side are retail-centric datacenters that generate real “neighborhood effects” by offering digital on-ramps to the major cloud players (AWS, Azure, Facebook, Google, Salesforce, etc.). In this segment, companies like EQIX and INXN in Europe dominate.  On the other side are wholesale colocation providers, a segment where low prices and rapid construction timelines are key to winning large deals from the hyperscale players… think CONE.

In a span of a year, IRM spent $1.8b to acquire its way into the datacenter space.  This was funded by debt and has put them at the high end of their historical leverage range at 5.6x.  


May 2018: Evoswitch ($235mm)

Dec 2017: IO Datacenters ($1.34b)

Oct 2017: Credit Suisse internal datacenters ($100mm)

Jul 2017: FORTRUST ($130mm)

The problem with these acquisitions is that they aren’t large enough to create scale on a national or a regional level.  The Company has 14 datacenters in US, Europe, and Asia and ~$230mm of TTM revenue. It does not have enough geographic coverage to attract major cloud players or to be able to build network that would then attract others into their datacenters.  

Sensing that, it seems that IRM is trying to capture some of the big hyperscale deals. It has begun construction on a 60MW “hyperscale-ready” Phoenix expansion.  The problem with building speculatively for hyperscale is that most hyperscale companies that lease from 3rd party datacenter companies work with them on build-to-suit deployments.  The best example is MSFT with CONE, where they have a great working relationship as CONE has proven they can build faster and cheaper than MSFT can. We think building a hyperscale-sized datacenter or spec is very risky.

Interestingly, there’s been turnover in the IRM datacenter division that we feel could impact execution.  VP of Portfolio of Operations, Data Centers Dana Adams, who was quoted regarding the IO acquisition ( left in Sept 2018 to become COO of AirTrunk, a datacenter company in Australia.


c) Dividend coverage eats up 88% of FCF, forcing IRM to issue debt and equity to fund datacenter requirements

With 100MW of capacity leased at 91.4%, future growth will have to come from developing new and adjacent datacenters.  This will require significant capital, as the Company has realized and started laying out in their slides.

Note that in FY2018, the dividend took up almost all the FCF of the business and that the Company needed to tap the debt and equity markets for the remaining $574mm.  In FY2019, they need an incremental $480mm, of which they’ve determined they can sell some RE to fund $100mm. The rest will likely come from equity issuances given the high leverage.  On $10b market cap, that would be 4% dilution.


d) Poor FCF conversion

The same slide regarding expected 2018 sources and uses of cash, when presented in 3Q18, showed $100-160mm of FCF available after dividends (see image below).  However, as you see above, the actual came in at $88mm. That’s despite already having 9 months in the books. EBITDA came in at the low end of the range, and the other big variance was in “customer inducements, relationships, and other.”  We’ll talk more about this line item in the next section.


e) EBITDA a better valuation methodology than AFFO

AFFO can probably be manipulated even more than EBITDA.  Each REIT has a different definition, but at its core it’s supposed to estimate the free cash flow that a REIT generates after taking into account recurring capex required to maintain its properties and revenue stream.  

What happens in the case of a business like IRM where volumes are declining?  Given that sales guidance also includes the impact of “customer inducements and acquisition of customer relationships,” should those uses of cash be included in the calculation of AFFO?  Even with $90-95mm of spend from essentially paying the breakage fees of customers to come over (much as AT&T would pay for you to break your contract with VZ), organic storage revenue growth is guided to be 1.75-2.5%, the lowest it’s been in a non-recession year.  This is despite really ramping up customer inducements in 4Q18 (addressed in prior bullet).

AFFO also doesn’t take into account the capital required to pivot away from the stagnant core business.

We look at valuation on an EBITDA basis.  As shown below, we see downside to $28/share based on 16x datacenter EBITDA and 10x legacy EBITDA to take into account the lack of growth and poorer than expected FCF conversion characteristics.


I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise do not hold a material investment in the issuer's securities.


  • Volume declines accelerate
  • Datacenter business does not scale as expected
  • Rates (positive or negative given it's a REIT)
    show   sort by    
      Back to top