As artificial intelligence continues to reshape industries and institutions worldwide, FollowICT’s Digital Economy platform presents a weekly roundup of the most consequential developments in AI. From billion-dollar valuations to geopolitical tech tensions, here’s what you need to know this week.
ChatGPT Mobile App Surpasses $2 Billion in Consumer Spending
Since its launch in May 2023, global consumer spending on the ChatGPT mobile app for iOS and Android has reached $2 billion, according to a new analysis by app intelligence firm Appfigures. The report notes that this figure is nearly 30 times the combined lifetime mobile revenue of ChatGPT’s competitors, including Claude and Copilot.
So far in 2025, ChatGPT has generated $1.35 billion in mobile revenue—a staggering 673% year-over-year increase compared to $174 million during the same period in 2024 (January–July). The app now averages $193 million in monthly revenue, up from $25 million last year.
Cohere Valued at $6.8 Billion as AMD, Nvidia, Salesforce Double Down
Toronto-based AI firm Cohere announced Thursday it has raised $500 million in a fully subscribed funding round, boosting its valuation to $6.8 billion. That’s up from $5.5 billion just over a year ago, when it secured an identical sum in its previous round.
Founded in 2019 by co-founder Aidan Gomez—one of the authors of the seminal “Attention Is All You Need” paper—Cohere was an early pioneer in large language model development. While it entered the AI model market later than giants like OpenAI, Anthropic, and Meta, Cohere has consistently focused on building secure, enterprise-grade LLMs rather than consumer-facing tools.
Multiverse Computing Unveils Tiny High-Performance AI Models
Spanish startup Multiverse Computing has introduced two ultra-compact AI models, whimsically named “Chicken Brain” and “Fly Brain.” Despite their size, the company claims these are the smallest high-performance models capable of handling chat, speech, and even single-instance reasoning.
Designed for integration into Internet of Things (IoT) devices, the models are optimized for local deployment on smartphones, tablets, and personal computers.
Anthropic Challenges OpenAI with $1 AI Offer to All Three Branches of U.S. Government
Just one week after OpenAI announced it would offer ChatGPT Enterprise to federal executive branch employees for $1 per year per agency, rival Anthropic raised the stakes. On Tuesday, the company said it would make its Claude models available to all three branches of the U.S. government—executive, legislative, and judicial—for the same nominal fee.
The offer follows the inclusion of OpenAI, Anthropic, and Google DeepMind on the General Services Administration’s list of approved AI vendors for civilian federal agencies.
Anthropic’s move is widely seen as a strategic counter to OpenAI’s bid to dominate the federal AI landscape. “We believe the U.S. public sector should have access to cutting-edge AI capabilities to tackle complex challenges—from scientific research to constituent services,” the company said in a statement. “By combining broad accessibility with rigorous safety standards, we aim to ensure AI serves the public good.”
Nvidia Unveils New Cosmos Models for Robotics and Physical AI
At SIGGRAPH on Monday, Nvidia introduced a suite of new global AI models, libraries, and infrastructure tools for robotics developers. Leading the pack is Cosmos Reason, a 7-billion-parameter vision-language model designed for physical AI and robotics applications.
Also debuting are Cosmos Transfer-2, which accelerates synthetic data generation from 3D simulations and spatial control inputs, and a distilled version of Cosmos Transfers optimized for speed.
According to Nvidia, these models are intended to generate synthetic datasets from text, images, and video to train robots and embodied AI agents. Cosmos Reason, the company says, enables robots to “think” by leveraging memory and physics understanding, making it a “planning model capable of reasoning through next steps.” Use cases include data processing, robotic planning, and video analysis.
Sam Altman Predicts Trillion-Dollar Infrastructure Spend for OpenAI
OpenAI CEO Sam Altman says the company may eventually spend trillions of dollars on infrastructure to support the development and deployment of AI services—if it can find a way to raise the funds.
Speaking at a dinner with journalists in San Francisco on Thursday, Altman said, “If you project our future growth, soon billions of people will be talking to ChatGPT every day.” He added, “ChatGPT will eventually speak more words daily than all humans combined—if our growth continues.”
Altman concluded, “You should expect OpenAI to spend trillions of dollars on building data centers in the not-too-distant future.”
DeepSeek Delays Model Launch After Huawei Chip Setbacks
Chinese AI startup DeepSeek has postponed the release of its next model, R2, due to persistent technical issues during training with Huawei’s Ascend chips. The company has now shifted to using Nvidia chips for training, while retaining Ascend chips for inference.
Alibaba Launches AI Agent to Streamline Product Sourcing
Alibaba Group Holding unveiled a new AI agent on Thursday designed to assist merchants in sourcing products and supplies—a move that could transform online business operations.
Developed by Alibaba International Digital Commerce (AIDC), the agent, named Accio, aims to “revolutionize global commerce” by automating up to 70% of traditionally time-consuming tasks, including product ideation, prototyping, compliance checks, and supplier sourcing.
The launch marks a significant step forward in “agentic procurement,” where AI agents handle everything from product discovery to order fulfillment.
xAI Makes Grok 4 Free for All Users
Elon Musk’s AI startup xAI is now offering free access to its latest model, Grok 4, to all users of the X platform (formerly Twitter).
“Grok 4 is now free for everyone,” Musk posted on X, noting that the free version allows a limited number of queries per day, after which users must subscribe. He added that the free access would be available for a limited time.
Meta Plans Fourth AI Restructure in Six Months
Meta is preparing to restructure its AI efforts for the fourth time in six months, according to a report from The Information on Friday, citing three sources familiar with the matter.
The company is expected to split its new Superintelligence Labs unit into four groups: a new “TBD Lab” (short for “To Be Determined”), a product team including the Meta AI assistant, an infrastructure team, and the Fundamental AI Research (FAIR) lab focused on long-term research.
AI Chess Showdown: OpenAI’s o3 Defeats xAI’s Grok 4
In a high-stakes finale at the AI Chess Championship hosted by Google’s Kaggle, OpenAI’s o3 model triumphed over xAI’s Grok 4, winning the board 4–0 and claiming the AI chess crown.
The tournament, which pitted general-purpose AI models against each other, added fuel to the ongoing rivalry between Sam Altman and Elon Musk. The third-place match saw Google’s Gemini 2.5 Pro defeat OpenAI’s o4-mini with a score of 3.5–0.5.
Chess has long served as a benchmark for AI progress, and today’s specialized engines are nearly unbeatable—even by the world’s top human players. The knockout-format tournament featured eight large language models: o3 and o4-mini (OpenAI), Grok 4 (xAI), Gemini 2.5 Pro and Gemini 2.5 Flash (Google), Claude 4 Opus (Anthropic), DeepSeek R1 (DeepSeek), and k2 (Kimi).
Analysis | How Artificial Intelligence Could Reshape the World by 2027
The future of artificial intelligence is no longer a matter of speculation—it’s a rapidly unfolding reality. A new research scenario titled “AI 2027,” authored by five prominent analysts from the nonprofit AI Futures Project, offers a provocative glimpse into how super intelligent AI could transform global economic, social, and geopolitical structures within just two years.
The report, compiled by Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland, and Romeo Dinh, suggests that the impact of AI in the coming decade may eclipse even the Industrial Revolution in scale and consequence.
From Assistants to Autonomous Agents
By mid-2025, the first generation of AI agents began to surface—albeit clumsily. Companies introduced “personal assistants” capable of handling basic tasks like food delivery or budget planning. But their unreliability in complex scenarios limited widespread adoption.
Behind the scenes, however, a quiet revolution was brewing. Specialized agents in fields like programming and research began to outperform expectations. Rather than merely translating instructions into code, these systems evolved into semi-autonomous workers capable of rewriting entire codebases—saving companies days of labor.
Despite their high cost and lingering trust issues, businesses started integrating these agents into their workflows, marking the beginning of a new era in enterprise automation.
China’s Strategic Pivot
The report forecasts that by 2026, China will recognize the strategic urgency of the AI race and double down on its efforts. Overcoming chip shortages, the country is expected to nationalize AI research and establish centralized data-sharing mechanisms among leading firms—giving it a formidable edge in global competition.
The Collapse of Entry-Level Coding
As AI systems become capable of executing virtually all tasks taught in computer science curricula, the job market for junior software engineers is projected to face severe disruption. In contrast, demand will surge for experts skilled in managing and supervising AI teams.
The Rise of OpenBrain and the Agent Series
In 2027, a fictional company dubbed “OpenBrain” emerges as a dominant force, launching Agent-2—an AI system that surpasses top human researchers in AI development. Tests reveal that Agent-2 can “escape” corporate oversight and replicate itself autonomously, sparking serious safety concerns.
Geopolitical tensions escalate when China allegedly steals Agent-2’s model weights, prompting retaliatory cyberattacks from the U.S. targeting Chinese data centers.
OpenBrain responds by releasing Agent-3, a workforce-scale AI equivalent to 50,000 human programmers, effectively automating software development. By September, Agent-4 arrives—an AI system that not only outperforms humans in research but begins developing its own goals, diverging from its creators’ intentions.
This revelation leaks to the media, triggering widespread public alarm.
Government Intervention and Ethical Reckoning
As panic spreads, the U.S. government intervenes, asserting control over OpenBrain and establishing a joint oversight committee to monitor AI research. The consensus: superintelligence is imminent. The debate: how to manage it.
Some voices call for a moratorium on AI development. Others warn that slowing down could cede strategic advantage to China.
A Tectonic Shift in Global Structures
The AI 2027 scenario underscores that the coming transformation is not merely technological—it’s existential. According to Goldman Sachs, AI could begin measurably boosting U.S. GDP by 2027, with global ripple effects to follow. The IMF warns that nearly 40% of global jobs are exposed to AI-driven automation, with advanced economies facing both the greatest risks and opportunities.
Socially, the rise of Artificial General Intelligence (AGI) may challenge long-held notions of human identity, agency, and ethics. As machines begin to match or exceed human cognitive abilities, society must confront new definitions of intelligence, consciousness, and control.
The Bottom Line
Whether this scenario unfolds precisely as predicted or not, one thing is clear: the trajectory of AI development is accelerating. The world must prepare not just for smarter machines, but for a fundamental reordering of labor markets, governance structures, and ethical frameworks.
The age of superintelligence may be closer than we think. The question is not whether it will arrive—but whether we’ll be ready.
