2 years ago
February 25, 2019
Appen Is Now Valued at USD 1.75 Billion as Investors Cheer 2018 Results
Australia’s Appen is riding high on booming markets for anything related to artificial intelligence. The Sydney-listed company, which supplies human-annotated datasets for machine learning and artificial intelligence to technology companies and governments, blew past expectations when it posted full-year 2018 results on February 25, 2019. Investors loved the results, sending Appen shares up 22%.
Appen grew revenues by 119% to AUD 364.3 (USD 260.9m) versus the prior year. Underlying EBITDA, which excludes M&A-related transaction costs, rose 153% to AUD 71.3 (USD 51.1m).
The key driver for growth was Appen’s November 2017 acquisition of US rival Leapforce, whose integration is now “all but complete,” said Appen CEO Mark Brayan.
China is fast emerging as an AI leader. According to Appen, it is actually the largest AI market outside the US. The company said they are making a “judicious investment” in China and are building out a team, customer base, and operations in Shanghai.
Appen said it employs 513 full-time employees and, similar to language service providers (LSPs), relies heavily on freelancers (over one million, according to the company). But unlike LSPs, Appen’s freelancers — who work in roles such as Search Engine Evaluators — typically do not require the level of qualification expected of professional linguists. “It’s not like we have to find certified translators,” CEO Brayan told Slator in a 2016 interview.
Appen operates two divisions called Content Relevance and Language Resources. Post Leapforce acquisition, Content Relevance now generates the bulk of revenues (AUD 312.8m in FY 2018). The unit helps improve search engine relevance algorithms with training data and by evaluating specific search results, entire search engine result pages, multimedia, maps, and news functions. Major clients include Google and Microsoft.
Appen’s smaller Language Resources division, meanwhile, organically grew revenues by 27% in FY 2018. The division has expanded its offering since the 2016 Slator interview and now provides annotated speech, natural language, and image data used in speech and image recognizers, machine translation, speech synthesizers, and other machine-learning technologies resulting in more engaging and fluent devices including Internet-connected devices, in-car automotive systems, and speech-enabled consumer electronics (according to company information).
Slator 2019 Neural Machine Translation Report: Deploying NMT in Operations
While Language Resources revenues grew strongly in 2018, EBITDA margins tightened from 30.1% to 21.8% due to a “significant reduction in complex, value added government work.”
Appen said its “successful expansion of the sales pipeline into technology sector” is mitigating the issue and has led to an increase in divisional H2 2018 revenue and earnings. Big tech’s drive to acquire language data has been a boon for a number of niche players.Moreover, some of the biggest LSPs have begun to highlight this line of business much more aggressively. Lionbridge CEO John Fennelly told Slator in a January 2019 interview about the LSP’s acquisition of Gengo and that he “can envision a day when our AI business is actually larger than our localization business.
This shift in messaging is understandable given Appen shares’ stellar performance. Since the 2015 IPO, Appen shares are up a staggering 4,480%. Appen’s market cap now stands at AUD 2.44bn (USD 1.74bn), passing RWS’ USD 1.65bn — although a bit of an apple-and-orange comparison given that the two companies operate in very different markets. Perhaps lured by Appen’s rise, Lionbridge’s private equity investors are reportedly working on a listing in far-flung Sydney.
Finally, Appen provided strong guidance for 2019 and projects FY 2019 EBITDA to be in the range of AUD 85–90m. The company said revenue plus orders in hand for delivery in 2019 already stood at AUD 165m at mid-February 2019.
Your information will never be shared with third parties. No Spam.