AI is transforming software companies

How large AI investments are changing the equity game for software stocks

Expanding the free cash flow framework

It has been a while since the last publication due to a long holiday. I have hinted at for some time, that the introduction of free cash flows as the most important concept in equity investing was to be expanded.

Today, we will explore the concepts a bit deeper and apply it to a proxy example mimicking Microsoft in a few ways. Specifically, the example will dive into the valuation mechanics of what higher capital expenditures (CapEx) mean for valuation of software companies given Microsoft’s latest announcement to expand CapEx to USD 80 billion in the fiscal year of 2025 (ends 30 June 2025). This is 80% more than FY24 and 235% more than FY22 which was the last fiscal year period before OpenAI announced ChatGPT in November 2022.

Some of these things will be a bit technical and advanced as some of the concepts have not been introduced beforehand, but rest assured, I think it will make sense anyway. The different underlying concepts will get more attention in future publications.

Incremental investments

As I explained in my previous publication, incremental investments are a part of the value drivers and thus key to equity valuation. The two primary investments a company makes are investments in net working capital (trade receivables, inventory, payables etc.) and fixed capital which is property, plant, and equipment (PPE). In the world of intangibles (software, patents, branding etc.), R&D and marketing can synthetically be capitalized and thus transformed into investments adding to the total CapEX of a company. This is a subject for another day.

Incremental fixed capital investments are investments less depreciation of previous investments. The incremental fixed capital rate is thus the net investments relative to the change in revenue. In other words, what degree of net investments are the company undertaking to drive revenue.

When Microsoft increases CapEx to USD 80 billion from USD 44.5 billion, the incremental fixed capital investment is USD 35.5 billion excluding the increase in depreciation. That is a sizeable change with profound implications for Microsoft's equity valuation, excluding any offsetting value drivers (think revenue growth).

The incremental investments relative to additional revenue is the key metric and value driver. Essentially the metric is calculated as:

\(\mathrm{Incremental~fixed~capital~rate}=\frac{(\text{CapEx}_{\text{t}+1} - \text{depreciation}_{\text{t}+1})}{(\text{revenue}_{\text{t}+1} - \text{revenue}_{\text{t}_0})}\)

As the chart above shows, Microsoft has significantly increased its incremental fixed capital rate between the pre-AI and post-AI periods¹. The incremental fixed capital rate was 34% in the fiscal years before the launch of ChatGPT and jumped to 125% and 88% in the subsequent two fiscal years. With Microsoft’s announcement of USD 80 billion in CapEx for the current fiscal year and my estimate of USD 100 billion in the following fiscal year the incremental fixed capital rate will increase to 247% and 186% in FY25 and FY26.

Intuitively, you understand that this has, all else being equal, negative implications for Microsoft and its valuation. As the recent years have shown the market has actually rewarded Microsoft for its decision to significantly increase its investments. Let’s examine at how it works.

Source: Yahoo Finance

How to get from FCF to corporate value

In order to understand how equity valuations work when a company significantly increases its investments, we must look at an example. I have created one that is quite similar to Microsoft.

We are assuming that we are looking at a software company that grows revenue by 13% each year and has an extraordinary operating profit margin of 40%. We also assume a low cash tax rate and few investments in working capital and a more moderate CapEx rate of 34%. The three other assumptions are WACC (weighted average cost of capital), long-term growth rate in NOPAT (used in the continuing value calculation), and finally RONIC (return on new invested capital).

The example excludes non-operating assets (interest bearing debt, other assets, excess cash, investment securities etc.). Essentially we are modelling this as a fully equity financed company. The WACC is therefore equal to the cost of equity which is the risk-free rate (the US 10-year yield) plus the firm specific equity risk premium (again a concept we will come back to). I have used figures from Aswath Damodaran to calculate the WACC.

In order to get to a corporate value that is identical to the observed market value we must forecast FCF 10 years into the future - not unrealistic. The required 10-year forecasting horizon to match the observed market value is also called the “competitive advantage period”.

Based on these numbers, this software company is valued at a FCF yield of 2.3% (334 / 14,268), which is quite similar to what we observe in Microsoft’s shares today.

Source: Peter Garnry

The corporate value of a company is essentially the present value of free cash flows during the forecasting period and then the present value of the continuing business after the forecasting period. The mechanics of discounting these cash flows and the methodology of calculating the continuing value will be discussed in a later publication. For now, just follow along and accept the math. The conclusions are intuitive regardless.

The impact on equity valuations from higher investments

Okay, so what happens when we change just one variable, the fixed capital rate, in the equity valuation machine? We increase the fixed capital rate to 150% from 34%. As you can see the free cash flows in the forecasting period go down and the corporate value actually declines by 12%. You're likely thinking that's not a lot, given the significant increase in fixed capital investments, right?

It is because the majority of this company’s value (or Microsoft for that matter) comes from the continuing value (74% to be exact) which lies after the forecasting period and is calculated off the NOPAT and not the FCF. I will explain later why that is the preferred methodology. What is important to understand here is that this high operating margin company is not suffering that much from a significant increase in fixed capital investments.

Source: Peter Garnry

So how come that Microsoft’s share price has actually increased then when a significant change in investments is negative? It is because these investments are expected to lift the revenue growth rate of the business. If I change the expected revenue growth rate from 13% per year to 15% then suddenly the corporate value goes up. The market expects Microsoft’s addressable market in the future to be much larger after ChatGPT relative to the period before. Two fiscal years have also gone by with Microsoft expanding revenue and operating margins, which helps increase its share price.

AI is making software companies more capital intensive

The reason why this is important is because these new AI systems have been the driving force behind the “Magnificent Seven” and the outrageous rise in equity index concentration (the highest in almost a 100 years). It is all driven by higher growth expectations for the software industry and productivity growth overall in the economy.

These new AI systems are transforming the software industry to be much more capital intensive than in the past. It is not a problem if we just model higher investments for an extended period and assume growth will, in fact, be higher. The danger lies in the fact that all capital intensive industries have lower operating margins and also a lower ROIC. What happens if we go back to our example and adjust the profitability lower due to the industry being more capital intensive?

The RONIC (essentially the ROIC after the forecasting period) is lowered to 15% from previously 25% (still above the cost of capital, so still driving shareholder value) driven by a lower operating margin (I set this to a high 30% but down from 40%). If I make these changes on top of the higher investments then the corporate value is 47% lower! To offset this impact I need to change the revenue growth rate to 23% from 13%.

This is essentially the fundamental question to think deep about when it comes to investing in the AI transformation. Will AI make the software industry more capital intensive and thus likely a lower operating margin business? If that is the case then equity valuations are very sensitive to a different narrative than the current one which is only focused on the growth component.

You have probably noticed that I frequently refer to concepts that I will explain later. There is a good reason for that. There are many concepts in equity valuation with the key book on valuation from McKinsey is around 800 pages long. We will just have to take things one at a time.

The next publication

The next publication will be less heavy on numbers and about football (the European version), my other passion. I will explain how long-term thinking, permanent capital, and the endowment philosophy can radically change football forever. It is not just a radical thought experiment as I am actually executing the idea in real life. Stay tuned.

1. The pre-AI and post-AI period is defined as before and after the introduction of ChatGPT on November 30, 2022. The use of ‘AI’ is more for understanding and use of a general term, because essentially it is before and after the commercialization of LLMs.