Concerns about AI-generated content are mitigated by AI detection tools, preserving data integrity and maintaining Reddit's value as a training data source.
Reddit's (NYSE:RDDT) shares have been on a slight decline since I last wrote on them, trailing the modest 5% gains in the broader market. Despite the negative price action, I think the company continues to be an excellent play as they collaborate with other tech firms to take advantage of the AI wave through their unique data set. As I covered in my last piece, Reddit has partnered with OpenAI earlier this year to leverage their vast repository of user-generated content, and lay the foundation to develop AI-driven tools to attract more advertising revenue. I think it's a holistic strategy that'll support Reddit's plan to integrate their content into OpenAI's ChatGPT platform.
Reddit did not stop there, and has also expanded their partnership with Sprinklr (CXM) by integrating Reddit's Data API with the software company's analytics capabilities. This enables Sprinklr to offer their enterprise customers deeper insights into consumer behavior and sentiment via Reddit forms.
I think we are seeing a pattern here: the launch of Reddit's Ads API is projected to create powerful new revenue streams and optimize the platform's advertising capabilities by allowing their partners & advertisers to integrate with Reddit's ad manager and build customized campaigns and analytics.
The company's monthly active user growth over the last year indicates they are well-positioned in an increasingly AI-driven, winner-takes-all social media environment. Given the strategic partnerships and Reddit's ability to harness AI to enhance both user experience and business analytics, I think the company remains a strong buy, despite the recent minor dip in share price.
Since my last analysis in May, where I emphasized Reddit's partnership with OpenAI, the company's stock has underperformed relative to the wider market. I think this has been due to the market questioning some smaller AI plays. I think in the case of Reddit, this is misguided.
Despite this, I think that the market has not yet fully recognized the value of their partnerships with OpenAI and Sprinklr. Reddit is ready to build an AI-driven future of digital media through its user engagement via their AI ads generator, and the monetization of their massive data trove. I think Reddit is still in a strong position to benefit.
The purpose of my follow-up coverage is intended to show how Reddit's most recent earnings report aligns with and even exceeds the expectations set in my previous write-up. The company continues to record strong growth in key metrics, such as user engagement and data monetization, which are directly supported by Reddit's strategic AI partnerships (like Open AI), and has now opened new revenue streams through AI-driven features and advertising.
I am confident about the company's ability to leverage their vast repository of human-generated content. While the broader market may be fixated on the competition among tech giants to develop the best AI models, Reddit's approach of providing data and integrating AI solutions is a smarter approach, in my opinion. The earnings report reinforces Reddit's potential to shape the future of AI-enhanced platforms.
Reddit's most recent earnings shows really strong performance in both user engagement and revenue growth. Key metrics from Q2 show that Reddit clocked a 51% year-over-year increase in Daily Active Uniques (DAUq), reaching an all-time high of 91.2 million users. This growth was driven by both U.S. and international markets, with U.S. DAUq increasing by 59% and international DAUq growing by 44% year-over-year.
Reddit's Weekly Active Uniques (WAUq) also saw excellent growth, jumping by 57% year-over-year to 342.3 million users. I think this surge in user activity shows Reddit's ability to bring in and retain a highly engaged audience, which was a critical factor in driving the platform's advertising revenue to grow by 41% year-over-year to $253.1 million in this same quarter.
Part of the secret internationally for growth were Reddit's Machine learning models that powered content translation, bringing more content from other languages into a users' native language so that they can interact with it. The shareholder letter noted that this was rolled out in France, but this is set to expand as noted on the earnings call:
International growth was a key driver in Q2, both in users and revenue. International daily active users exceeded 45 million, growing 44% year-over-year and 11% from Q1. This was led by focus countries like France, India, the U.K. and the Philippines, where we're seeing good results with our growth strategies, including through machine translation.
Immersive=machine translation is now widely available in French, and as a result, France was one of our fastest growing countries surpassing growth in the U.S. We will now begin to expand machine translation to German, Spanish and Portuguese, aiming to make Reddit accessible to everyone regardless of their native language. Looking ahead, we're making progress across our emerging initiatives, including the user economy, search and data licensing.
As this rolls out outside the US, the benefit of this is that non-English-speaking countries will be able to benefit from English content. I expect the vice versa to happen soon for US users as well as non-English content on key topics is brought back to US users. This will help make the platform exponentially more valuable to users of all origins. This should help continue to power growth (like the exponential growth they are seeing in France).
On top of Machine learning models for growth, Reddit is leveraging advanced AI models to help enhance their advertising offering, which mirrors the approach by Meta (META). Through their partnership with Google (GOOG) (GOOGL), the company integrates AI-driven tools that allow advertisers to build more targeted ad campaigns on the platform. This collaboration, valued at $60 million, grants Google access to Reddit's vast repository of user-generated content to train its AI models on. In return, Reddit gains access to Google's sophisticated LLMs, which are being deployed to help improve the platform's internal search features and ad offerings for creatives. With better ad creation tools, Reddit believes creators will have a higher willingness to pay for Reddit ads since they will be more curated and targeted. I agree.
Reddit's approach to AI in advertising is designed to enhance the effectiveness of their ads through machine learning algorithms that analyze user behavior and preferences to best serve ad content. Advertisers can then create even more personalized and impactful campaigns, similar to how Meta uses AI to optimize ad delivery across its platforms. Advertisers use tools that enable deeper engagement with their target audiences to drive higher conversion rates and better return on ad spend.
Through this compelling value proposition to advertisers, Reddit's integration of AI models helps diversify their revenue streams beyond traditional advertising. The company recently acquired Memorable AI to help integrate AI-driven intelligence tools and ad creative into their offerings for advertisers. This acquisition adds foundational AI models to Reddit's advertising stack, with the goal of optimizing ad creative and beginning to see the impact of this integration next year platform-wide.
The company's straightforward AI licensing deals, including those with AI companies like Google and Open AI, are also expected to generate strong income.
CEO Steve Huffman said during the Q2 earnings call:
The number of comments viewed reached a new high, up 10% from Q1 to Q2. Over the years, we've hosted some notable Ask Me Anything sessions, ranging from U.S. presidents to celebrities and sports figures to regular people with interesting stories. We introduced several updates to the AMA feature, making it easier for a host to schedule and post and easier for users to participate.
During the pilot, we've seen hundreds of new AMAs created, including those with high-profile brands and artists. These product updates tie into our advertising strategy, allowing us to test new conversation ad placements and AMA ads, which are high-intent services for brands and businesses to reach their audience.
As the platform continues to scale, so does our need for safety and moderation tools. Using AI, we are reducing the moderation burden and making Reddit's platform safer at scale.
Reddit's revenue growth for multiple upcoming fiscal years are expected to well outpace the sector median (this is key for later). For this fiscal year ending this December, the company is expected to generate $1.19 billion in revenue. From here, revenue is projected to reach $1.46 billion the following year (2025), jumping by 22.59%.
From a valuation perspective, Reddit's shares are trading at a higher price-to-sales multiple relative to the broader sector. The company's forward price-to-sales ratio clocks in at 7.82 (against the sector median of 1.27), which, while elevated, I think shows the market is starting to value Reddit's capacity to monetize their strong and accelerating user base.
To illustrate the potential upside, if Reddit's price-to-sales multiple were to rise to 10.5 (which I think it should, given the blockbuster growth they are having and their AI initiatives), this (assuming a revenue estimate of $1.19 billion for FY2024 and a multiple of 10.5) would imply a market cap closer to $12.5 billion. This translates to an upside of approximately 34.26% from current levels.
For Reddit, like any content platform, one of the biggest risks is the challenge of controlling the amount of AI-generated content that permeates their platform. Since Reddit is a repository of freely accessible user-generated content, it inadvertently becomes a training ground for large language models to learn from human behavior (this is what Open AI is paying Reddit for access to do).
Ironically, this creates a paradox where AI systems might begin to train on data that they themselves have accidentally generated (i.e. Open AI trains on a post that it doesn't realize one of its models made), leading to a phenomenon known as "model collapse."
AI models, over time, become less accurate and more homogenous in their outputs due to the recycling of AI-generated data.
When AI models are trained on data that is partially or wholly AI-generated, it often leads to a feedback loop where errors and biases in the model are compounded rather than corrected. Ironically, it's like a narcissist listening to their own beliefs over and over again. If a narcissist becomes engulfed in their own beliefs, they lose the flexibility to learn from others and become a better person. AI models are the same way.
For LLMs, this can cause overall system performance to severely degrade, even as individual models improve, according to one study.
To mitigate this risk (and keep selling their content as training data to Open AI and others), Reddit can employ AI content detection tools, which are now widely available, to internally monitor and flag AI-generated content so it's not fed to Open AI (or others) and diminish the value of the data that Reddit's AI customers train off of. I think this will help preserve the integrity of the data as a whole on Reddit, and also help maintain the quality of the AI models that rely on this data for training.
Reddit's recent performance, while underperforming the broader market, fails to reflect how strongly the company remains positioned for powerful future growth. Strategic partnerships with OpenAI and Sprinklr continue to improve their AI-driven offerings and help drive higher user engagement (a strong feedback loop). Despite concerns about AI-generated content on the platform and the spillover risks of self-defeating improvements to AI models by training on this content, Reddit's unique ability to leverage their immense (and compounding) user base continues to make them a compelling investment opportunity, in my opinion. With this, I think shares remain a Strong Buy. Social media is seen as a mature market in the US. I think Reddit will prove this notion wrong.
