A Historic Week in AI

A Historic Week in AI

Last week marked one of the greatest weeks in AI since OpenAI unveiled ChatGPT causing turmoil in the markets and uncertainty across Silicon Valley. 



DeepSeek R1 makes Silicon Valley quiver. 

Shockwaves hit Silicon Valley last week when Liang Wenfeng and the Chinese hedge fund High Flyers launched their open-source LLM model DeepSeek-R1.

 

 

  • A model understood to be at least as good as premium versions of OpenAI's latest O3 model and purportedly trained for only $5.58 million - A fraction of the training costs of comparable models. 
  •  
  • NVIDIA stock crashed 18% on the news as DeepSeek changes the narrative that effective AI needs unlimited GPU's and venture capital to train more intelligent models. Silicon Valley AI executive compensation is also in the spotlight as many of the big names take home more than it cost to train to the Chinese LLM. 

    DeepSeek have also adopted Meta LLama's strategy as a late entrant and made their model open-source available to download and run offline, further distancing itself from (Not)OpenAI and endearing surprised respect from the AI community.

    There are however concerns from users that DeepMind records keystrokes and other sensitive information in servers in mainland China and murmurings from industry leaders like Sam Altman OpenAI's technology may have been copied. 

    Source: https://www.deepseek.com/

     
  • OpenAI Operator

    OpenAI themselves had major news, unveiling Operator. A major breakthrough in the usability of AI agents and agentic workflows.

 

  •  
  • Operator is autonomous AI agent that can control your browser and complete web based tasks such as online booking, shopping and pretty much anything a human can do with a keyboard and mouse. 


Operator unfortunately is only available to US users on the Pro plan ($200) however there are some free alternative versions such as Open Operator and Browser-Use.

 

Source: https://openai.com/index/introducing-operator/

 


Gemini 2.0 Flash Thinking

 

Last month Google unveiled a substantial upgrade to its AI model, Gemini 2.0 which they claim is “built for the agentic era”, where AI can understand, plan, and take actions on behalf of users. This new model builds on the success of Gemini 1.0, offering enhanced multimodal capabilities (text, images, audio, video) and native tool use, enabling more advanced AI agents. 

 


Google DeepMind have now released Gemini 2.0 Flash Thinking, a significant update to its AI model, offering advanced reasoning, transparency, and performance improvements which are trending towards parity with premium versions of AI.

Google is also increasingly integrating AI into its Workspace tools, offering features like real-time document assistance and automated summarization to streamline workflows for users.

In a rapid response to this announcement, OpenAI introduced new features, including memory capabilities for ChatGPT and finalizing the O3 model. This move was seen as an attempt to maintain its competitive edge in the rapidly evolving AI space1.

Source: https://deepmind.google/technologies/gemini/flash-thinking/

 

 

  • Trump's Stargate

 

President Donald Trump announced the ambitious initiative "Stargate" on January 21, 2025, which involves a joint venture between OpenAI, Oracle, SoftBank, and MGX. 

 

The project aims to invest at least $500 billion in AI infrastructure within the United States over the next four years and described as the largest AI infrastructure project in history, with immediate funding of $100 billion set to be deployed.

 

 

The primary goal of Stargate is to construct a “colossal data center” to support the advancement of artificial intelligence, specifically aiming to enhance U.S. competitiveness against nations like China in the AI sector. Trump emphasized that this project would lead to the creation of over 100,000 American jobs "almost immediately."

 

Key players involved include OpenAI CEO Sam Altman, SoftBank CEO Masayoshi Son, Oracle Chairman Larry Ellison, and representatives from MGX, with SoftBank taking financial responsibility and OpenAI overseeing operations. The venture also features technology partnerships with Microsoft, ARM, and NVIDIA.

 

Discover AI Agents