These days we are all talking about Artificial Intelligence (AI) and machine learning. But do you know how they actually come about? Or how companies actually build them? Well, this leads to our Appen stock forecast we will be reviewing today. We will see why this australian company, that’s involved in AI, still has substantial value waiting to be unlocked.
The company’s key activities are in collecting, classifying, translating, reviewing and labeling large volumes of data in the forms of image, text, speech, audio, video and other categories used to build and train AI systems.
On the surface, you might think that this data business is boring. It does not seem as exciting as companies launching AI-related technologies and applications.
But these data collections and transformations may in fact be the most important component of creating smart technologies.
To appreciate the value that Appen is providing, you need to have some ideas of AI and machine learning.
Appen Stock Forecast: Understanding the Importance of Data in AI
Here we will explain the workings and essence of AI and machine learning. So that you can appreciate the value that Appen is doing.
To begin, AI is the broader concept here. And we usually understand it as more of an automation, from softwares to physical machines. Key idea is that these technologies will be able to perform tasks that usually require human intelligence.
It covers a huge range of applications. For instance, AI will return more relevant results in your search engines. On the other hand, AI gives autonomous cars its self-driving capabilities.
So creating these technologies and systems actually boils down to building a good model. That’s can only happen if you provide quality data to train and build the model. This will then allow computer systems to predict the likely outcomes that humans will perform with high accuracy.
Appen Stock Forecast: Understanding Machine Learning
To understand this, let’s go back to our mathematics in secondary school.
First, recall one of the most basic graphs: y = mx + c. For instance, consider y = 2x + 3.
By feeding the variable x with a certain number, say 5, it can translate to a y value of 13.
However, in the real world, we are usually dealing with systems that don’t follow such a perfect relationship. Which is why we also learn this thing called predictive model.
Best Fit Curve VS Data Training
Recall the time when you were doing your science experiment. Or remember during one of your secondary mathematics class. You are probably asked to draw a best-fit line from a bunch of data points collected from your experiment.
Source: Abacus / Reporting Statistical Results in Your Paper
Take this case study of the relationship between the dry mass of waterlilies and the number of seeds they contain.
The relationship between these 2 variables are recorded based on the experimental observations and measurements made.
By drawing the best-fit line, you can now predict the number of seeds that a particular waterlily with certain dry mass will contain.
In fact, nowadays we have computers to help us draw such a best-fit line or curve relatively quickly.
Source: jmp / Fitting the Multiple Linear Regression Model
The idea is to use various mathematical equations and statistics to minimize the error or difference between the predicted y values and the actual recorded y values, based on each input x variable. By solving these collectively, it will help us find the equation of the best fit line model.
In this case, it will be y = 1.779x – 35.84.
As observed from the chart, this model passes through all the experimental values pretty well.
Now you have a working model to predict the likely number of seeds, base on a particular waterlily’s dry mass.
Appen Stock Forecast: Data Integrity
But imagine that the data collected on the initial dry mass and corresponding amount of seeds are subject to errors. Perhaps due to inaccuracies or varying experimental conditions. Then the mathematical model you built will not be as effective or accurate.
This is in fact the key operating concepts behind AI. Data scientists are using machine learning to use collected data to train and produce mathematical models for useful applications.
Of course, this x and y linear relationship we have discussed here is a simplified model and scenario. We have many more complex mathematical models and methodologies when it comes to modeling and predicting real life scenarios.
Source: Analytics Vidhya / Commonly used machine Learning Algorithms
For instance, above are just commonly used machine learning algorithms used to train and build AI systems.
Of course we will not go into the details of all these technical terms.
The objective here is to show you the importance of high quality data in building AI technologies. Because garbage in, garbage out. Whatever quality of data you used to feed the algorithms, you can expect equivalent quality of prediction with your model.
Now you can see why Appen’s data business could potentially be more important than what you would have thought initially. It is the brain behind the actual softwares or applications you could be using.
Appen Stock Forecast: Business Overview
With a better understanding of Appen’s data business in the AI industry, let’s look at the company’s operation in detail.
Source: Appen Annual Report FY2020 / Operating segment information
The company segregates its business into two key operating segments – Relevance and Speech & Image.
For the Relevance segment, it offers annotated training data used as input source to improve the performance of search engines, social media, and e-commerce applications.
As for the Speech & Image segment, it offers training data used to build AI-based voice interface, translation, text-analysis, AR/VR, and image perception systems.
Observe that the Relevance segment takes up 90% of the services revenue. Whereas the Speech & Image segment makes up the remaining 10%.
Source: Appen Annual Report FY2020 / Disaggregation of services revenue
Also, note that the Relevance segment is solely based on its business in the U.S. Its Speech & Image operation is concentrated in Australia, with some business in the U.S and other countries.
Appen’s major customers include Google, Facebook and Amazon. The company did not give a detailed breakdown. But there’s indication that its revenue is still mainly driven by these big technology firms.
Appen Stock Forecast: Performance Review
Source: Appen Results Presentation FY2020 / Solid growth maintained in FY20
Appen saw a 12% increase in its topline to A$599.9 million in FY2020. This is driven by its business in the Relevance segment, which saw its revenue climb 15%.
The Relevance operation benefited from increased demand for data annotation in existing and new projects. Usual higher revenue level in Q4 was not seen in FY2020. That’s due to its major customers’ shifting priorities with the coronavirus pandemic situation.
However, existing customers continue to require the annotated data to train, update and refresh their AI models. This is required to ensure that their models stay relevant and free from bias.
On the other hand, the Speech & Image revenue dropped by 10%. This is due to project cancellations and delays in relation to the Covid-19 situation. Also, the pandemic has negatively affected Appen’s ability to work on growing their project pipeline and winning new customers.
Source: Appen Annual Report FY2020 / Financial Highlights
Overall, the company’s revenue for FY2020 was not spared from the impact of the pandemic.
Although its revenue did rise, it is at a much slower pace compared to the past. The revenue in the past few years recorded a compounded annual growth rate (CAGR) of 49% from 2015 – 2019.
Similarly, from 2015 – 2019, its underlying EBITDA and underlying Earnings Per Share (EPS) grew at a CAGR of 51% and 44% respectively.
So, compared to the previous years, definitely the market is disappointed with the company’s results in FY2020.
Appen Stock Forecast: Market Reaction
Source: Yahoo Finance / Appen Limited Share Price Chart
As a result, the market has penalized its share price. It plummeted from a high of more than A$40 in Aug 2020, to the current A$11 level.
However, at this level, I think Appen is undervalued at the moment.
Growth and Valuation
Source: PR Newswire / Data Collection And Labeling Market Size Worth $8.22 Billion By 2020
The data collection and labeling market is forecasted to grow at a CAGR of more than 25% from 2021 to 2028.
Source: PR Newswire / Insights on the Global Artificial Intelligence (AI) Market 2021 – 2025
Similarly, the AI market in general is forecasted to grow at a CAGR of more than 20% from 2021 to 2025.
Appen is one of the leading data collection and labeling players in the market. Hence, it should grow at least with the industry pace of 25% annually over the next few years.
To be conservative, let’s use the growth rate of the general AI market. Meaning, let’s use 20% as the company’s growth rate instead.
Source: Appen Annual Report FY2020 / Consolidated statement of profit or loss and other comprehensive income – Earnings Per Share
Take Appen’s diluted EPS of A$0.4085. And assume that its earnings will grow at 20% per annum for the next 5 years. Thus, its EPS will increase to about A$1.02.
Now assume that its Price-to-Earnings (P/E) ratio equilibrate from the current 27 to a lower figure of 20.
Its target share price should hit A$20.40, almost two times its current value of A$11 a share.
Apart from that, we see some positive elements and outlook for the company. These support our optimism for the company moving forward.
Appen Stock Forecast: Positive Elements and Outlook
Strong Balance Sheet
Source: Appen Results Presentation FY2020 / Solid growth maintained in FY20
As of the end of 2020, Appen recorded A$78 million of cash in its balance sheet. It has no debt at the moment. Such a healthy balance sheet will enable the company to ride through any potential crisis in the future. This is definitely a plus point for its shareholders.
Customer and Project Wins
Source: Appen Results Presentation FY2020 / New customer and project wins
Next, we saw that the company is continuously adding new customers to its pipeline. In 2020 itself, it added 136 new customers across multiple industries and geographies.
Not only that, it has managed to secure a 34% increase in projects with existing major customers. Yes, these new projects tend to be smaller in scale. But increasing data types and use cases will create a tighter integration, and result in greater retention of customers.
Rapid Growth In China
Source: Appen Results Presentation FY2020 / China Revenue
Appen’s business in China represents a very small contribution to an almost A$600 million overall company’s topline in 2020. But it is growing at a very rapid pace.
In fact, the company’s China business revenue grew 60% quarter-on-quarter in 2020.
Appen Stock Forecast: High Retention Rate and Growth From New Customers
Source: Appen Results Presentation FY2020 / Growth from new and long-standing customers
Now, perhaps the most important metric to look at.
At first I was thinking that such data provision service is just a one-off business. So, it may not be ideal in terms of long-term sustainability of its business.
However, take a look at the chart. The revenue contributions from the earlier cohort of customers, represented by the darker color tones, are growing higher and higher, at least for its key operating segment in Relevance.
This is because its major programs are highly retentive. As discussed earlier, Appen’s core clients will require the annotated data to train, update and refresh their AI models constantly. This is to ensure that their models stay relevant and free of bias.
As a result, we observed continued growth in the service revenue from major customers.
Furthermore, the company is also actively adding new customers each year to diversify its customer base. This will help to boost its topline and reduce reliance on a few key customers in the future.
Therefore, this retention chart gives me the confidence of the stickiness and the competitive advantage of its business. Such barriers of entry will enable Appen sustain its growth in the future.
Appen Stock Forecast: Growing Global Crowd Workforce
Source: Appen Results Presentation FY2020 / Creating value for our stakeholders – Global crowd
Finally, I want to mention Appen’s growing global crowd workforce.
This represents the skilled and diverse crowd of over 1 million plus contractors, living in more than 170 countries and speaking 235 languages.
The global crowd gives the company its key ability to serve their customers. They provide Appen the flexibility and ability to scale up and down according to customers’ needs for their projects.
Not only that, the diversity of its global crowd is critical in providing quality and real-world applicability of the training data.
This growing base of contractors will create a network effect. The company can tap on its community of contractors to grow their business further. Something that may be difficult to replicate for new entrants.
Let’s sum up this Appen stock forecast. The company is positioned to ride on the waves of the AI technologies in the years to come.
As a leading data collection and labeling service provider, it has created trust and relationships with key technology companies to complement their service offering ranging from search engines to social media and ecommerce.
We feel that the company does exhibit competitive advantages in terms of its huge yet flexible global contractors crowd, as well as the growing databases that they are building on.
Yes, there’s risks of intensifying competitions and pressure from key customers re-prioritizing its AI projects moving ahead. Also, not forgetting the pandemic situation that may continue to affect its business.
Source: Yahoo Finance / Appen Limited Stock Summary
However, there’s heavy market correction to the current valuation. Given its relatively strong business positioning, I think there’s more room for Appen to grow. Its share price could potentially hit A$20 again in the future.
With the tailwind of the growing market for AI, I think the company’s prospective upside outweighs the potential downside risks.
Definitely adding Appen to my portfolio.
Let me know your thoughts on this, and share with us in the comment below. And remember to subscribe to our newsletter below. So that you can be informed once our next investing tip is out.
In the meantime, check out other insights and analyses we have done. Continue learning so that you make better investment decisions.
Keep learning and happy investing. 🙂