What was once a darling among Aussie tech investors is now a shell of its former glory. Falling from atop its AI spire, shares in the training and data labelling company are now 41.5% below where they finished last year. This is made all the more confusing by the marvellous moves of other global companies with exposure to the booming sector.
Though, how much further can such a monumental decline continue? Especially given Appen's relevance in what is considered one of the hottest industries right now.
Could the market's persistent scalding now be misplaced?
The crowd is in question
Appen's first-half results for 2023 did not inspire confidence among investors. Facing a challenging environment, the data solutions company posted a 24% reduction in revenue, falling to US$138.9 million compared to the prior corresponding period.
Furthermore, Appen's underlying EBITDA suffered during the period, tumbling into a loss for the half. The reduced revenue profile and increased costs were attributed as drivers of the negative result.
These facts alone wouldn't necessarily be an issue for a long-term investor if they represented a blip in a generally positive trend. Unfortunately, analysts and shareholders have grown more suspicious of whether Appen could face an existential threat — a fear reflected in the Appen share price.
As an example, one analyst on Appen's earnings call remarked:
Has there been any consideration that maybe the crowd model is dead? Maybe there's a disruption somewhere that Appen is not able to take advantage of… and we're just working with rebranding the old stuff?
What is being referenced here is Appen's reliance on paying people to provide data labelling to create a training dataset. This requires real people to be paid to undertake the manually intensive task of observing and annotating data.
However, Appen dismisses that human involvement is outdated. Instead, CEO Armughan Ahmad references what is called 'reinforcement learning from human feedback (RLHF)' as an ongoing need for Appen's services. Even so, not all the research comes to the same definitive conclusion.
In March this year, MIT Technology Review reported on an interesting finding by AI safety and research company Anthopic. According to the non-peer-reviewed paper, some large language models had shown evidence their outputs could be corrected for biases simply by prompting them to do so.
At the least, it may cast doubt on the future demand for Appen's services.
What do the fundamentals say about the Appen share price?
Appen currently holds a market capitalisation of $216 million, which roughly equates to a price-to-book (P/B) value of 1. However, it's hard to say whether the company's assets are truly worth their currently quoted value if they were to be sold.
On a positive note, Appen held a net cash balance of US$55 million with no debt at the end of June. This may suggest the company is at little risk of any immediate financial strain.
Despite this, Peter Day of Sequoia Wealth Management still is not a buyer of Appen at this share price, as published in The Bull. Day states that challenges remain while customers examine their AI strategies.