Revenue spread, page-4

@Paccman,

• In CY2020, "global services" revenue (aka Appen's major US technology customers) generated $328.1M USD compared to total revenue of $413M USD, equating to ~79% of the total. This is trending in the right direction with customer diversity increasing and revenue concentration decreasing over the past 5 years of listed history.
• Appen's major US technology customers comprise 8 out of the top 10 US technology giants. These include Facebook, Amazon, Google, and Microsoft. The majority of this revenue is comprised of content relevance work with Microsoft and Facebook.
• Historically, in 2015, Appen's business was heavily dependant on Microsoft with approximately 50% of total revenues stemming from MSFT alone. Then Facebook joined the fold around 2017 and at that point in time, the two giants together were estimated to generate circa 75% of Appen's total revenue.
• More recently, on the other side of the world, Appen claims to be working with the major Asian technology giants: BATTSS – Baidu, Alibaba, Tencent, TSMC, Softbank & Samsung. This will provide revenue diversification over time if Appen maintains its early foothold and market share here.
• In my view, there is a customer concentration risk at play with Appen, albeit, this risk is lessened because there are multiple projects with each major customer, lessening dependency on any one piece of work. In 2018, Mark Brayan brushed off criticisms the company had a concentrated customer base, asserting that the business could keep growing quickly even if it doesn't win another major customer.
• In terms of automotives, Scale AI appears to have had some success with western automotive companies, such as GM, Toyota and Lift, but Appen appears to have the best reach across the Asian automotive companies, claiming to work with most of China's automotives.
• At this point in time, it appears that Tesla are not working with any of the data labelling providers (Appen, Lionbridge, Scale AI etc), one might assume that this is all in house functionality currently and a competitive edge of Tesla's since they have a bank of millions of miles driven to date, with early mover advantage in the EV and AV space. Tesla is also angling to have less dependency on LIDAR, which is the training data type that players like Appen can assist with.
• Read more: https://techcrunch.com/sponsor/appen/building-world-class-ai-for-autonomous-vehicles-requires-a-single-source-of-reliable-data/

Hope that helps,

T.E.P.