The artificial intelligence (AI) realm witnesses rapid developments around the world, reshaping the world of digital transformation and accelerating the pace of AI adoption. In this weekly newsletter, FollowICT is highlighting the most important news and advancements in AI around the globe.
Nvidia in Talks to Acquire Lepton AI
Reports indicate that the semiconductor giant is close to finalizing a deal to acquire Lepton AI, a company that rents servers powered by Nvidia’s AI chips, according to The Information. Citing anonymous sources, the site reported that the deal is valued in the hundreds of millions of dollars.
Founded two years ago, Lepton AI received $11 million in seed funding in May 2023 from CRV and Fusion Fund. Another major player in the server rental market is Together AI, a startup that has raised over $500 million in venture capital, although it is only about a year ahead of Lepton, according to TechCrunch.
Delay in New AI Image Feature in ChatGPT for Free Users
OpenAI CEO Sam Altman announced on Wednesday that the launch of new AI image features in ChatGPT for free users will be delayed, citing demand much higher than the company anticipated.
Altman stated in a post on X on Wednesday, “Images in ChatGPT are much more popular than we expected (and we had very high expectations).”
The company announced on Tuesday the launch of the original image creation feature GPT-4o, allowing users to upload and modify images, noting that it will soon be available for all tiers of ChatGPT. As of Wednesday, this feature was only rolled out to ChatGPT Pro, Plus, and Teams subscribers.
Quora’s Poe App Launches Subscription Plan for $5 per Month**
The Poe app, Quora’s chatbot application, launched one of the most affordable subscription options on Tuesday, at just $5 per month.
Additionally, the company introduced a higher-priced plan at $250 per month, designed for users who need to send a large number of messages on Poe.
Poe allows users to utilize multiple AI-powered bots—including DeepSeek-R1, GPT-4o, Claude 3.7 Sonnet, o3-mini, ElevenLabs, and others—in one place. The system operates on a points system, enabling users to spend points on different models, with each bot having its own point cost per message.
Google Begins Rolling Out Screen Sharing and Live Video Features in “Gemini”
A Google spokesperson stated that the company has started rolling out new AI features in Gemini that allow the smart assistant to see what is happening on the user’s phone screen or through the phone’s camera and respond to questions about either instantly.
These features come nearly a year after Google first showcased “Project Astra,” which powers these capabilities.
The live video feature enables the smart assistant “Gemini” to interpret a stream from the smartphone camera in real-time and answer questions about it.
SoftBank Plans to Support AI Efforts in America with $1 tln
Japanese conglomerate SoftBank plans to create AI industrial complexes across the United States and is considering an investment exceeding $1 trillion.
According to the Nikkei newspaper, Masayoshi Son, the founder and CEO of the group, is expected to visit the US to discuss his ideas regarding these industrial complexes.
The factories are likely to use AI-powered robots that operate independently without human supervision due to labor shortages in the country.
Son collaborated with “OpenAI” and “Oracle” in January to unveil a $100 billion joint project to fund AI infrastructure in the United States, one of the first commitments of its kind following Donald Trump’s presidency.
German Company n8n Raises $60 mln for AI-Driven Workflow Automation
Developer tools are rapidly changing with AI. Companies facilitating AI adoption in workflows are seeing increased interest. After a startup named n8n (pronounced “enay-ten”) revamped its workflow automation platform to be more AI-friendly in 2022, the company reported a fivefold increase in revenue, doubling in just the past two months.
In light of this growth, TechCrunch confirmed that n8n raised €55 million ($60 million) in funding, and the Berlin-based company announced it now has over 3,000 business clients and around 200,000 active registered users. The startup will use this second funding round to continue investing in technology and expand into new markets, such as the U.S., which accounts for more than half of n8n’s user base. The company does not disclose its revenue, and the client count includes both free and paid users, as well as subscribers to short- and long-term subscriptions.
Microsoft Enhances “Copilot” with Smart, Advanced Search Tools
Microsoft is launching a deep search tool powered by AI in its “Microsoft 365 Copilot” application, which is an intelligent chatbot application.
A recent suite of deep search tools is being launched across chatbots, including “ChatGPT” from “OpenAI,” “Gemini” from “Google,” and “Groq” from “xAI.”
These tools operate on what are called logical AI models, which possess the ability to think through problems and validate them—skills that are arguably crucial for conducting in-depth research on a topic.
Microsoft’s versions are called “Researcher” and “Analyst,” with the “Researcher” version combining deep search capabilities from “OpenAI” with advanced formatting and deep search potentials.
Chinese Company DeepSeek Updates Its AI Model “V3”
DeepSeek has released updates to its AI model “V3,” promising to deliver better programming capabilities, confirming the Chinese startup’s commitment to staying a step ahead of its leading American competitors like “OpenAI” and “Anthropic.”
The “V3-0324” update, which was released last week on the AI development platform “Hugging Face” without an official announcement, is said to address real-world challenges while setting standards for accuracy and efficiency, according to several reports.
“V3” is an older model from DeepSeek launched last December, followed by the release of its “R1” model.
New AI-Powered Weather Forecasts Available Across Africa
Google has announced the availability of new short-term rainfall forecasts across Africa via its search engine. This enhances its efforts to use machine learning to tackle the challenge of predicting rainfall in the coming hours.
These forecasts have become possible thanks to advancements in Google Research’s MetNet model for real-time weather predictions (very short-term weather forecasts), which utilizes satellite data and ground observations to produce advanced rainfall forecasts in data-scarce regions worldwide. Google’s technology can predict global rainfall with high accuracy within a five-kilometer radius every 15 minutes for the next 12 hours—all in under a minute.
-Google Unveils Next Generation of AI Thinking Models
On Tuesday, Google revealed Gemini 2.5, a new series of AI models that pause to think before answering a question.
To launch this new series of models, Google is releasing Gemini 2.5 Pro Experimental, a multimodal AI model that the company claims is its smartest model yet.
This model will be available on Google’s developer platform, Google AI Studio, as well as in the Gemini app for subscribers to the $20 per month Gemini Advanced AI package. In the future, Google states that all new AI models will include built-in reasoning capabilities.
-Alibaba Launches AI Model Capable of Processing Video and Audio on Phone
Alibaba Group has launched a new AI model within its “Qwen” series, which the company claims can process text, images, audio, and video, and is efficient enough to run directly on smartphones and laptops.
The company added that the new model is expected to be used to build what are known as AI agents that can assist visually impaired individuals in navigating their environments through instantaneous audio descriptions. The new model, launched on Wednesday, is called “Qwen2.5-Omni-7B.”
The new model is multimodal, meaning it can process inputs including text, images, audio, and video, while generating natural instant text and audio responses.
Alibaba has made the new model open-source on the “Hugging Face” and “GitHub” platforms, following a growing trend in China towards open-source models, inspired by DeepSeek, which made its leading “R1” model open-source.
PwC Launches New Platform to Help AI Agents Work Together
PwC has been building and deploying AI agents for various clients over the past nineteen months, working independently. During that time, management noticed that the agents often operate like “ships passing in the night,” according to Matt Wood, PwC’s global and U.S. chief technology and commercial innovation officer, in a statement to Business Insider.
PwC does not believe that silent communication will succeed in the age of AI. Wood explained that their goal is to “transform the ships from mere vessels passing at night to a harmonious fleet.”
On Thursday, PwC unveiled the “agent OS,” a new platform it describes as an “operating dashboard” for enterprise AI. This platform allows companies to build, customize, and connect agents to automate complex tasks.
PwC stated in a press release that the Agent OS is designed for a world where companies develop AI agents in various ways, whether embedding them as features within platforms, as standalone applications, or as highly specialized agents built on proprietary or open-source software.
The company stated that the Agent OS integrates with various systems, from Anthropic to Google Cloud and Microsoft Azure. Wood explained that users can test workflows directly in the Agent OS and integrate them into the systems their teams or clients are already working on.
Thus, a marketing executive at a retail company can accelerate a new advertising campaign by connecting creative generation agents built through OpenAI with testing agents on Google Cloud and analytics agents from Salesforce.
PwC stated that once the platform is operational in a client environment, it can deploy multi-agent solutions in just two weeks.
First AI Factory in Africa Announced with Expansion Plans to Egypt
On 24 March, Cassava Technologies, owned by Zimbabwean billionaire Strive Masiyiwa, announced a partnership with Nvidia to build the first AI factory in Africa. This is not just an ordinary data center; it is a specialized power hub designed specifically for AI computing.
Cassava plans to use Nvidia’s advanced AI technology in South Africa by June 2025, with plans to expand to Egypt, Kenya, Morocco, and Nigeria. The company has not specified a timeline for this.
The facility operates with Nvidia’s supercomputers utilizing GPUs. It will provide what Cassava calls “AI as a Service” across the continent, leveraging a vast fiber-optic network and energy-efficient data centers.