Google Moves More Aggressively into the AI Market for Lawyers

Google has launched Gemini Enterprise for Legal, a specialized version of its AI platform designed for law firms and legal departments. The system includes AI agents and capabilities for tasks such as contract analysis, legal research, citation verification, monitoring regulatory changes, and other workflows involving confidential data. The platform also connects with legal products and databases already used by professionals, including services associated with Thomson Reuters and other companies in the legal-tech ecosystem.

Why it matters: Enterprise AI is beginning to specialize. In fields such as law, it is not enough for a model to generate a good answer; the system must also respect confidentiality, access rights, internal policies, and the need to ground information in legal sources. Google is therefore betting on the next stage of the market: not simply “AI for everyone,” but infrastructure designed for professions where context and control are just as important as model performance.

Apple Bets on AI Running Locally, Not Just in the Cloud

Apple has introduced new versions of the Mac mini and Mac Studio, positioning them at a time when an increasing number of developers and users are using Mac computers to run AI models and agents locally. The new systems offer more powerful processors and memory configurations designed in part for AI workloads, while interest in running AI agents directly on users' own hardware is becoming an increasingly important part of the market.

Why it matters: Until now, the AI boom has been associated primarily with massive data centers. Growing interest in local AI could move part of the computing process closer to the user, offering potential advantages in data control, latency, and independence from cloud services. This does not mean data centers are disappearing, but rather that the AI market is beginning to split between extremely large models running centrally and models or agents capable of operating directly on users' devices.

China's Race for AI Chips: Enflame Prepares an IPO of Nearly $900 Million

Chinese AI chipmaker Enflame Technology will open subscriptions on September 2 for an initial public offering on Shanghai's STAR Market, aiming to raise approximately 6 billion yuan, or around $892 million. The company, backed by Tencent, is part of a group of Chinese manufacturers seeking to build domestic alternatives for the hardware required by AI. The funds are expected to support the development and commercialization of new generations of chips and closer integration between hardware and software.

Why it matters: AI competition is not taking place only between GPT, Gemini, Claude, and other models. Behind them is a battle for chips, energy, factories, and computing capacity. Enflame's IPO is another sign that China is trying to build a domestic industry capable of reducing its dependence on Western technology. In AI, control over physical infrastructure could become just as important as software performance.

Anthropic Sees a Potential AI Market Worth More Than $30 Trillion

According to information published by The Wall Street Journal, Anthropic estimates its total addressable market at more than $30 trillion. The figure is expected to be presented to investors as the company prepares for a potential public offering and reflects an extremely broad definition of economic activities that could be performed or automated with the help of AI.

Why it matters: The $30 trillion figure is not a guaranteed revenue forecast, but rather the theoretical size of the market Anthropic believes it could address. Still, the number illustrates the scale of the bet being made by the AI industry: major AI labs are no longer competing only for the existing software market, but are assuming that artificial intelligence will take over part of the work currently carried out by people and traditional systems. If that assumption proves correct, AI will not simply become a new industry, but a technology capable of reshaping the scale of industries that already exist.

Claude Begins Remembering Projects Across Conversations and Cowork

Anthropic has unified the memory used in Claude conversations with that of Claude Cowork. Users can therefore move from discussion and research to executing a task without having to explain the project's context all over again. The company also allows users to view, edit, and delete information stored in memory, while certain categories of sensitive information are not retained by default.

Why it matters: Memory is one of the key components that could transform a chatbot into a persistent assistant. An AI system that already knows a user's project, preferences, and context can move more easily from “tell me something” to “do something for me.” Yet this evolution also makes control over stored information increasingly important: as AI becomes more useful through context, users need to understand more clearly what the system retains and how they can control that context.

A German Broker Opens Investments to ChatGPT and Claude

German broker Scalable Capital has announced that its users can use AI chatbots such as ChatGPT and Claude to analyze portfolios and execute trades through the platform. The company says the new service is a first step toward broader AI integration across its products and that access is accompanied by security measures. Scalable has more than one million customers and over €60 billion in assets.

Why it matters: This is one of the clearest examples of AI moving from providing information to taking action in a regulated sector. A chatbot that explains what is happening in the market is one thing; a system connected to an investment account is something else. As AI gains access to financial services, questions about authentication, action authorization, and accountability for decisions become just as important as the technology itself.

Primero Raises $12 Million to Bring AI to Major Companies in Latin America

Mexican startup Primero has emerged from stealth with a $12 million seed round led by Kaszek and General Catalyst. The company is trying to solve a common problem among large organizations in Latin America: legacy and fragmented software systems, where an important share of business rules and institutional knowledge still exists in employees' experience rather than in digital systems. Primero's platform aims to connect these systems and enable the deployment of AI agents capable of carrying out complex tasks.

Why it matters: Primero's story shows that the next stage of AI will not be decided exclusively in Silicon Valley. Many of the biggest opportunities may emerge in companies that do not need another chatbot, but rather a way to connect AI to the infrastructure they already use. If regional startups can understand the specific characteristics of local markets and integrate AI into legacy systems, they may build advantages that global platforms cannot easily replicate.

Mistral and HUMAIN Build a New AI Alliance in Saudi Arabia

French company Mistral AI and HUMAIN, the Saudi AI company, have announced a collaboration covering computing infrastructure, the development of advanced models, and the deployment of AI solutions in Saudi Arabia and across the region. The partnership includes plans to develop high-performance Arabic-language models and use data center infrastructure to expand local computing capacity. The value of the collaboration has been described as being in the hundreds of millions of euros.

Why it matters: There is a natural connection here with AIdapted's analysis of Mistral and the transformation of European AI sovereignty into a commercial product. While in Europe the French company has focused on building and securing computing capacity, its partnership with HUMAIN shows how the same logic can be extended into an international ecosystem. AI sovereignty no longer simply means having your own model; it also means controlling the language, data, infrastructure, and ability to adapt technology to the needs of a particular region.

Bucharest Enters a Global Ranking of Cities with Potential for AI Data Centers

Bucharest ranks 21st in the Savills Power and Place Index, a ranking that analyzes cities with the greatest potential for future data center development. The assessment takes into account factors such as energy availability, telecommunications infrastructure, access to resources, and the capacity to develop new projects. According to the analysis, the next wave of investment in digital infrastructure will be strongly influenced by growing demand generated by AI.

Why it matters: For Romania, this is a story about positioning rather than a data center that has already been built. As major European hubs face challenges related to energy, land availability, and grid connections, cities capable of offering energy capacity and infrastructure could become important alternatives. AIdapted has already examined how the competition for AI is shifting the focus from models toward infrastructure and sovereignty; in this context, Bucharest's position suggests that Romania could play a role not only as a user of AI, but also as part of the physical infrastructure supporting the industry.

Sources