Artificial intelligence and machine learning have become an integral component of everyday technology and this trend is set to become exponentially more pronounced in the coming years. Appen, as a 3rd party dataset provider for artificial intelligence (‘AI’) and machine learning algorithms, is set to be a key beneficiary of this structural change.

Since listing in 2015, Appen has established itself as a success story and market favourite. This is for good reason, as the company has consistently proven itself to be a cash-producing machine, with rapid top-line growth and reliable profitability becoming synonymous with the name.

Understanding Appen’s Business Operations

As a company operating in a new and potentially confusing industry, it’s worth taking a bit of time to understand the operations of Appen’s business and how that plays into the growth of the AI and machine learning industry.

Appen’s core operations are in the provision of what is referred to as ‘labelled datasets’ for companies developing AI and machine learning algorithms. A piece of data can come in many forms, such as an image or audio. Labelling the data simply means to describe what is contained in it, such as a dog in an image or the word “Hello” in an audio file. Data labelling is carried by Appen’s distributed online workforce, and these datasets are then provided in mass volume to Appen’s customers.

Labelled datasets are critical in developing an algorithm, as this is where the algorithm learns what is represented in an image, or what the words in a piece of audio represent. By feeding this “training data” into the algorithm, it can then detect this information in future pieces of data. 

The more volume of training data and the higher the quality of labelling being input into the algorithm, the better the quality of outputs it will generate as the machine is “optimised”. Conversely, low-quality labelling otherwise known as “bad data” will lead to poor quality machine learning models. This makes Appen’s services a business-critical function for any AI project team.

As machine learning continues to advance, the consequences of bad data grow exponentially. Systems that use deep learning neural networks require large volumes of accurate and complete data. This data must be well labelled to ensure the system creates generalisations which will correctly recognise future objects or words.

The two-year chart above is evidence of a successfully executed strategy by Appen’s management. Shareholders have been rewarded with a 3x return on their investment over the last two years. Appen’s earnings have not grown as fast as the share price, meaning there has been a re-rating in the company’s earnings multiple. However, as we discuss later, we believe this re-rating is justified and a near term earnings multiple is not the best means of evaluating Appen as an investment.

Impact of COVID-19

COVID-19 can impact a business like Appen in two ways. Pandemic responses such as social distancing requirements can have an impact on the business’s ability to operate efficiently, while a broader economic deterioration can decrease demand for its services.

Appen is in the enviable position of having its 1m+ ‘crowd’ of global workers already operating from home. Social distancing requirements did not involve any changes to the current operations of its crowd and like many other technology businesses, Appen has been able to transition its full-time employees to work from home with very little disruption to business operations.

Secondly, the broader economic impacts of this pandemic appear to have had very little impact on its end user demand. Appen provided a business update in mid-April. To date it has not seen any slowdown in digital advertising or IT spending and as such has reinstated its prior full year earnings guidance.

Where we Differ from the Market

We believe the market is underestimating the size and duration of the growth in demand for 3rd party data labelling services from the AI and machine learning industry. A common misconception when it comes to Appen is that it is easy for Appen’s clients and competitors to replicate its services. 

Although Appen may not have the internally developed IP and associated barriers of some technology companies, Appen’s high quality processes and scale make it difficult to efficiently replicate. Appen’s business-critical data engineering function also mean a low-cost alternative is typically not worth the associated risk.

What Are the Key Metrics We Look At

Appen is a highly cash generative business that is benefitting from exponential growth in demand for its products and services. While we expect the growth rate to moderate over time, the structural nature of the change in demand for its services means a step change in revenue is likely to be maintained for a least a decade. It is for this reason we believe Appen should not be valued using a near term earnings multiple.

The pertinent question when assessing Appen’s attractiveness from a valuation perspective is: does the current market valuation compensate an investor for the stream of cashflows that will be generated by the business over the next decade. In our view, we believe the share price significantly undervalues the present value of those future cashflows.

Should we Expect a Capital Raising?

Appen sits in a position of financial strength and stability. With cash resources in excess of A$100mn at the most recent update and highly cash generative operations, Appen appears highly unlikely to find itself in a position where it must raise capital. This is not to say that Appen will not raise capital if it identifies attractive opportunities. Historically the company has used acquisitions as a strategic tool. As an example, its acquisition of Leapforce in 2017 improved its capabilities in search relevance and enabled the business to increase its participation in the growth of AI.

Appen’s Leapforce acquisition was followed up in 2018 with the acquisition of Figure 8. Figure 8 provided it with a global leading software platform that delivers automated annotation tools for AI teams. This acquisition further increased Appen’s competitive advantage against other 3rd party data vendors and critically, diversified its revenue away from arguably higher risk project-based work. This strategically provided Appen with a stable recurring revenue base to build on.

The company stated in their most recent release that it “is well-positioned to weather the pandemic as well as respond to opportunities as they may arise.” While our base case does not forecast any capital raisings from Appen, we believe that accretive opportunities in current market conditions would be well received by investors.

To Summarise

We believe the investment case for Appen continues to stack up. Management has proven their execution skills and continues to execute their strategy at an exceptional level, establishing the company as a global leader in AI and machine learning datasets. This industry is set to compound at an extraordinary rate over the coming decade, further augmenting the company’s growth.

It’s rare to find a company with a combination of strong management, attractive structural tailwinds and a proven history of growth and execution at a reasonable price, but when we do, we like to be on the long side of the bet. We believe Appen demonstrates these characteristics in spades. 

Invest in the leaders of tomorrow

We look to exploit inefficiencies in the small and microcap segment of the market by identifying at an early stage those smaller companies that are the beneficiaries of change. Stay up to date with our latest insights by clicking ‘follow’ below, or ‘contact’ us for more information.