Tech On Me: SearchGPT prepares to launch, Outbrain snaps up Teads, Chinese tech at the Olympics, and much more

This week’s focus

Outbrain has confirmed its acquisition of Teads, the video monetisation company owned by European telco Altice. This deal, valued at approximately $1 billion, involves an initial cash payment of $725 million, a deferred $25 million, and 35 million shares of Outbrain stock worth around $169 million. Altice had sought to sell Teads to address its significant debt.

Upon completion of the acquisition, Outbrain CEO David Kostman will lead the combined entity, with Teads co-CEOs Bertrand Quesada and Jeremy Arditi stepping in as co-presidents. This merger is seen as a strategic move to create a comprehensive platform for the open internet, expanding Outbrain's reach into TV and video content through Teads' SSP capabilities.

My take: By integrating Teads' capabilities, Outbrain can tap into the lucrative CTV and video advertising markets, which are proliferating and attracting substantial ad spending. The acquisition is also a strategic response to the growing scrutiny and criticism of content recommendation platforms like Outbrain and Taboola. These platforms can be associated with lower-quality ad placements and MFA (made-for-advertising) content. 

By incorporating Teads' more premium offerings, Outbrain can improve its reputation and attract higher-paying advertisers. This move could also help the company distance itself from the negative perceptions of its previous business model and reposition itself as a provider of high-quality, brand-safe advertising environments.

Another aspect that stands out to me is the financial structuring of the deal. Outbrain's ability to secure commitments from major financial institutions like Goldman Sachs, Jefferies, and Mizuho Bank reflects confidence in the strategic rationale behind the acquisition. It also indicates that Outbrain is willing to leverage significant resources to ensure this deal goes through, highlighting the importance of this acquisition for its future growth and market positioning.

In other news

Is Singapore-based e-commerce platform Qoo10 in trouble?

Young-bae Ku, founder and CEO of Singapore-based marketplace Qoo10, has committed to using his personal assets to secure approximately $58 million to compensate South Korean customers and vendors affected by the company’s liquidity crisis. This crisis resulted in failed payments to about 60,000 small merchants on its Korean marketplace subsidiaries, TMON and WeMakePrice.

During an emergency hearing with the South Korean government, Ku, who owns a 42.77% stake in Qoo10, stated he would raise the funds within 30 days, potentially by selling his stake or using it as collateral. Due to payment failures, the hearing followed the court receivership filings of TMON and WeMakePrice.

Qoo10 announced plans to obtain emergency liquidity from overseas funds, sell subsidiaries, or use them as collateral. The company had previously informed Korean financial authorities of its intention to raise $50 million but had not submitted detailed plans.

My take: It is commendable that Young-bae Ku has pledged to use his assets to secure $58 million to address the liquidity crisis. This demonstrates a responsibility and commitment not always seen in corporate leaders. However, this situation also underscores the precariousness of rapid expansion through acquisitions. While acquiring companies like WeMakePrice and TMON may have initially seemed like strategic moves to strengthen Qoo10’s market position, they have placed significant financial strain on the company.

The liquidity crisis and subsequent fallout highlight the risks associated with aggressive growth strategies in the e-commerce sector. Substantial capital needed to support acquisitions and integrate new operations can quickly become a burden if not managed carefully. This situation might be a cautionary tale for e-commerce platforms like Alibaba and Shopee, which are pursuing similar growth trajectories.

Meta unveils video vision AI 

Meta has introduced Segment Anything 2 (SA2), an advanced version of its successful machine-learning model that extends its capabilities from images to video. This development, showcased by CEO Mark Zuckerberg at SIGGRAPH, highlights the rapid progress in computer vision. Segmentation refers to a model's ability to identify and outline different parts within an image, and SA2 enhances this by efficiently processing video content.

Zuckerberg emphasised the model's utility for scientific research, such as studying coral reefs and natural habitats. Despite the increased computational demands of processing video, SA2's performance demonstrates significant advancements in efficiency within the industry. Like its predecessor, SA2 will be open and accessible, with a demo available. Meta is also releasing an annotated database of 50,000 videos for training purposes, though another internal database of 100,000 videos used for the same purpose remains undisclosed.

My take: Meta's decision to extend its segmentation capabilities from images to video highlights a significant leap forward in computer vision. The ability to process video efficiently without overwhelming data centres is a testament to the progress in computational power and model optimisation.

From a practical standpoint, SA2's potential applications are vast and exciting. Researchers studying environmental changes, such as coral reefs, will find this tool invaluable. However, the implications extend far beyond academia. Media and security industries can leverage this technology for more sophisticated video analysis and content creation.

Chinese firms showcase innovative tech at Olympics 

Alibaba Group's cloud services distribute over two-thirds of the live broadcast signals, using their OBS Cloud 3.0 to aid media outlets in accessing images and videos. This cloud-based distribution method, replacing traditional satellite broadcasting, offers low latency, high resilience, scalability, and cost-efficiency benefits. In addition, Alibaba Cloud has deployed AI-enabled multicamera replay systems at 14 venues, providing slow-motion replays and 3D models for deeper performance analysis. SenseTime assists China's basketball team with AI-driven sports data analysis, offering real-time feedback to improve training and competition strategies.

My take: I am both impressed and contemplative about the broader implications. The strides made by firms like Alibaba and SenseTime are remarkable, showcasing the pace and scale of technological advancement in China. Integrating cloud computing and AI to enhance live broadcasts and athletic performance is a testament to their innovation and ambition—and adding extra value for brands who wish to leverage this tech.

However, this also raises questions about the broader impact of such technological dominance. While it's clear that Chinese companies are leading the charge in AI and cloud computing, their rapid progress and the scale at which they operate could overshadow smaller, less resourceful competitors. The emphasis on technological superiority and the race to outdo one another might also lead to ethical and privacy concerns, especially when dealing with vast data.

Reddit buys Memorable AI to boost ads offering

Reddit has acquired Memorable AI, a startup specialising in ad creative tailored to previous campaign performance metrics, to bolster its competitive edge against Google, Meta, Amazon, and TikTok. The acquisition includes Memorable AI’s entire team and is set to enhance Reddit's ad product performance by integrating creative intelligence with its existing contextual intelligence. This move allows Reddit to optimise creative content across various media types, including display and video. 

Memorable AI, backed by industry leaders like MediaLink’s Michael Kassan and Scope3’s Brian O’Kelley, has gained significant attention and a solid client base, including major brands like Unilever and ExxonMobil. Memorable AI helps advertisers save on creative testing costs, offering insights on optimising creative elements to drive desired metrics. 

Reddit will incorporate Memorable AI’s technology into its platform, providing advertisers data-driven feedback on their ad creatives. This technology will analyse past campaign data to suggest improvements and enhance ad performance without significant media spending.

My take: From my perspective, integrating Memorable AI's technology into Reddit's platform is a game-changer. The ability to tailor ad creatives based on past campaign performance metrics means that advertisers can optimise their content with a level of previously unattainable precision. 

This is particularly critical as creative optimisation becomes essential for driving ad performance in today's data-driven marketing environment. By incorporating these capabilities, Reddit can offer advertisers data-driven feedback on their ad creatives, enhancing performance without significant media spending.

Looking ahead

OpenAI recently launched SearchGPT, an AI-powered search engine that aims to revolutionise how users find real-time information online. 

Unlike traditional search engines, SearchGPT organises and summarises search results, providing concise and relevant responses. For instance, it can offer brief descriptions of events or detailed planting advice with proper source attribution. 

Currently, a prototype is available to 10,000 users. SearchGPT is powered by GPT-4 models and works with third-party partners for content curation. OpenAI plans to integrate these features into ChatGPT. This move potentially positions OpenAI against giants like Google and startups like Perplexity, but with a focus on collaboration and clear source attribution to avoid past criticisms of content misuse.

Check out this week's Sounding Board with APAC tech leaders on the above here, and more on their thoughts in the video below.