Is NXT the next compounding machine?

When thinking about businesses growing earnings does not necessarily mean you are growing shareholder value. The focus of management should be to invest now to maximise future cashflow and not focus on current earnings.

I think the 2004 Amazon shareholder letter from Jeff Bezos sums up this better than I could.

In this example, Bezos talks about the problem that high capex businesses have, they have to invest a large portion of capital upfront and then bring the asset up to full utilisation as quickly as they can. To invest further capital in the business to grow earnings would destroy shareholder value.

So in that context let’s run the ruler over NXT. NXT has built data centres in most Australian mainland capitals with a data capacity of 43MW and is currently in the process of expanding its data centres by an additional 61MW.

NXT is a high initial capex business where each MW of installed capacity can cost $8-$10m. So in the context of the 2004 Amazon shareholder letter has NXT achieved a high ROIC on its existing assets and is it justified in expanding the capital base?

Current Business Snapshot

The following table shows a quick summay of each data centre in the portfolio.There is a difference to the total average revenue MW figure I calculate at $3.71 vs the company at $3.99. I have simply divided the revenue by the Utilisation at the end of the period. Maybe the difference is reflected in average utilisation over the half.

It is apparent that B1 is the most mature and the most profitable centre in the portfolio. It averages revenue of $7 per MW for H12017 as compared to $3.83 for M1 and $2.72 for S1. This is obviously a function of the profit mix of the 2 centres vs whitespace/retail. B1 also has the highest utilisation rate followed closely by M1 and S1. Canberra and Perth have been the laggards in the portfolio.

The following chart shows the ROIC over time for the B1, M1 and S1. I have excluded P1 and C1 for now, but bear in mind these 2 assets have had a not insignificant amount of capital invested at $65m.

This ROIC is calculated on 30% tax rate and maintenance capex of 5% of revenue (ie equal to the D in EBITDA).

NXT is earning rates of return far in excess of its WACC, especially in Brisbane and Melbourne. So it makes sense to try and reinvest capital into these markets to extract similar returns on capital. This is Capital Allocation 101.

Currently I calculate the overall ROIC for the entire capital base around 17% if you annualise H12017 numbers.

It is a shame that NXT cannot allocate all its capital in Brisbane, go where the money is and go there often.

It is interesting that S1 has a lower return on capital than M1. This shows up in the average revenue per MW, where M1 achieves $3.83 vs S1 of $2.72. I am sure the company is aware of this dynamic but it would be interesting to note if this is competition driven rather than as a result of product mix.

The following tables provide a detailed breakdown of the profitability of each centre, B1, M1 and S1.

The key assumptions in calculating NOPAT are capex maintenance is 5% of revenue ( I believe Equinix in the US is quoted at 3%) and final tax rate of 30%

So what does that mean for cashflows?

Business value is driven by its present value of current cashflows plus present value of future cashflow it can earn on reinvesting capital in new business. The limitations of pinpoint valuations are well known as they require a lot of assumptions and assumptions can be massaged to give us the result we want.

Present value of current cashflow from current invested capital

To date NXT has invested around $290m in its existing datacentres. Due to timing of acquiring clients and the product mix, the return on capital has varied to around 23% for B1 to the less mature 9.6% for S1 as we showed in the previous tables.

If we assume that a blended variation of ROIC at 20% is an achievable number on current capital, then this would give an unlevered NOPAT of around $58m once the business mature. Currently this blended ROIC is around 17% so we are allowing for some more operational leverage in the existing business.

The following table shows a quick DCF model valuation of around $2.91 based on this NOPAT.Assumptions are discount rate of 10%, terminal growth in NOPAT is forecast to be 4% after the existing assets reach full maturity at a blended ROIC of 20%. I have not carried the debt on this portion of the valuation but rather carry it on the valuation for future creation.

It is worth pointing out the calculation of Terminal Value. I believe the following formula is appropriate as advocated by Michael Mauboussin et al.g = future terminal growth, RONIC = return on new invested capital.

On this basis the current business is worth $2.91. We will run a sensitivity table on different growth and discount rates later.

So what is future capital allocation worth?

The company has flagged approximately $350m to be invested in B2, M2 and S2 over the next couple of years. From previous experience it has taken around 5 years to fill out the facility.

If we assume a ROIC on a blended basis of 15% for this new capital (a reduction from Gen1) then this will result in NOPAT of $53m by the end of year 5. Again we are assuming an amount for maintenance capex of 5% of revenue and a tax rate of 30%. I have made no allowance for working capital as this at the margin. The ramp up to maximum ROIC I assumed is linear over the 5 years.Assumptions on terminal value for revenue growth at maturity is 4%, discount rate at 10% and capital is reinvested at 15%. Here the net debt is carried at $160, giving a value of $1.23 per share. Add this to the existing business and I calculate a total valuation of $4.14 per share.


I think a pin point valuation is not particularly helpful. A range of potential outcomes are more useful, as valuation is more art. The following tables show valuation around different growth rates and different discount rates.

The first table assumes the company can reinvest its capital at 15%. The columns shown different terminal growth rates with the rows showing different discount rates. For example terminal growth at 4% with a discount rate of 10% gives a valuation of $4.14.The second table shows similar changes but assumes the company can reinvest its capital at 20%.Conclusion

NXT is benefiting from a reduction in long term yields as NXT is perceived to be similar to an infrastructure or utility stock. I think this is an unfair comparison as my reading indicates utilities being heavily regulated average a ROIC in the US around 6.3% which is considerably lower than that achieved by NXT.

At current prices of around $4.60 I believe the market is predicting that NXT is going to grow earnings by 4-5% in the future after Gen 2 and be able to reinvest back in the business and achieve returns of 20% on this additional capital (discount at 10%).

Historically only businesses with a wide moat have been able to consistently reinvest their capital at these high rates. In the competitive landscape there are a number of large US players in the data centre space so economics would dictate that these outsized returns should be competed away over time.

There is no doubt that NXT has the potential to be a great compounder and is a quality business. However the market on these metrics seems to be pricing in these future expectations already.