Amazon is reportedly working on an AI model with advanced reasoning capabilities that could compete with OpenAI’s o3-mini and DeepSeek’s R1, among others. This newest model is expected to launch as early as June and under the Amazon Nova brand; one of Amazon’s reported goals with “Nova” is to provide a more cost-effective model than those of its competitors. In addition to affordability, Amazon aims to achieve a top-five ranking based on external performance benchmarks assessing software development and math skills.
This move underscores Amazon's commitment to building its own suite of AI models while also promoting a diverse range of options through its Bedrock platform. It is worth noting that the development of its own reasoning model places Amazon in direct competition with Anthropic, a company with which it has a close partnership, having invested $8 billion in the AI startup and collaborated on various projects, including AI chips and cloud computing.
More info at Business Insider and TechCrunch.
Meanwhile, Google has introduced a new experimental embedding model, “Gemini Embedding,” which is now available through the Gemini developer API. Embedding models convert text inputs, e.g. words and phrases, into numerical representations that capture their semantic meaning. As they can enhance performance while reducing costs and improving latency, these embeddings are widely used in applications like RAG, recommendations, and classification. While this is not the first embedding model offered by Google, Gemini Embedding is the first to be trained on the Gemini family of AI models, granting it a deeper understanding of language and context.
According to Google, Gemini Embedding outperforms its predecessor, text-embedding-004, and shows competitive results on popular embedding benchmarks. The new model can also handle larger text and code inputs and supports over 100 languages, doubling the capacity of its earlier version. Currently in an experimental phase with limited availability, Gemini Embedding could get a stable release date in the coming months, demonstrating Google’s commitment to its AI offerings.
Read more at TechCrunch.
The UK's Competition and Markets Authority (CMA) has announced that Microsoft's partnership with OpenAI would not warrant an investigation under the merger provisions of 2002’s Enterprise Act. The CMA concluded that Microsoft does not control OpenAI's commercial policy but rather exerts significant influence over it. This decision follows an investigation that began in December 2023, prompted by concerns that Microsoft's investment in OpenAI—nearly $14 billion since 2019—could hinder competition within the AI industry in Great Britain. The CMA noted that recent developments have diminished Microsoft's control over OpenAI, such as a recent renegotiation of their cloud computing agreement. Additionally, Microsoft opted out of taking a board seat at OpenAI – a decision that could have otherwise attracted more regulatory scrutiny. While it continues to monitor the tech industry's investments in AI, the CMA hasn't found evidence of wrongdoings here or in similar cases, including Google's dealings with OpenAI competitor Anthropic.
Meanwhile, OpenAI has announced the launch of NextGenAI. This brand new consortium brings together 15 leading research institutions with a common goal: utilizing AI to accelerate research and transform education. With a commitment of $50 million in various fundings, OpenAI seeks to foster collaboration to achieve breakthroughs more rapidly. According to OpenAI, NextGenAI is designed not only to fuel innovative discoveries but also to prepare the next generation to shape the future of AI. The founding partners of NextGenAI include institutions such as Caltech, Harvard University, and the University of Oxford, among others.
For more information, check OpenAI’s blog and TechCrunch.
Google has issued an official apology following a weekend-long disruption in its Google Ads service. Beginning on March 1, the outage resulted in some accounts not serving ads and others experiencing error messages, high latency, and various other issues. According to several reports, this outage has led to substantial financial losses for advertisers, with some estimating the impact to be in the tens of thousands of dollars. Although Google stated that the issues had been resolved and that only a "small number of accounts" were affected, many advertisers are still grappling with the consequences, including displaced ad spend and lower conversion rates.
Media buyers have reported varying degrees of impact, with some clients experiencing losses of up to €50,000 due to the blackout. Many advertisers expressed frustration over the lack of clear communication from Google regarding what caused the disruption and how many accounts were affected. As a result of the outage, impacted clients might have had to overspend to recover lost ground, risking further declines in conversion performance.
Read more at Adweek.
Discover all the latest news, articles, webinar replays and fifty-five events in our monthly newsletter, Tea O'Clock.