AI Digest: AI University in Astana, $965 Billion Anthropic, and Google's Digital Twin
Over the past two weeks, several significant events have taken place in the world of technology — from global AI company records to regional breakthroughs in education and language technologies. Let's look at the main highlights.
Freedom Telecom International Opens Laboratory in Silicon Valley
A joint laboratory by Freedom Telecom International has opened in Silicon Valley. The new facility will serve as the holding company's presence in one of the world's key technological ecosystems. The company notes that competition today is no longer between products, but between ecosystems and the speed of access to knowledge and technologies. The laboratory will enable faster integration of international practices into the holding company's operations.
Kazakhstan Launches First AI University — Qazaq AI Research University
Qazaq AI Research University, the first AI-enabled university in Central Asia, has opened in Astana. Artificial intelligence is embedded in the educational process: students are supported by 24/7 AI tutors that create individualized learning paths. The bachelor's program lasts three years, and instruction is conducted in English. Graduates complete their studies with a ready-made AI product in their portfolio. The campus is located on the EXPO grounds in Astana and is integrated with Alem.cloud and supercomputing infrastructure. Document submissions are open from June 20 to August 25 via eGov and the OQU platform.
Anthropic Closes $65 Billion Round at $965 Billion Valuation
Anthropic raised $65 billion in a Series H round at a post-money valuation of $965 billion, surpassing OpenAI to become the world's most valuable AI startup. The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, with participants including Samsung, SK Hynix, Micron, Blackstone, and Amazon, which had previously committed $5 billion.
The funds will be directed toward safety research, expanding computing power for Claude, and scaling products. This round may be the final one before an IPO.
GPT-5.5 Taught to Control Windows: AI Agent Codex Can Click and Type Text for You
OpenAI has expanded the capabilities of the Codex agent by adding support for the "computer use" feature on Windows. The system can now see screen content, move the cursor, click buttons, and type text—just as a human user would. Previously, this feature was available only on Mac. Codex performs actions in applications through the graphical interface, without manual command entry via the terminal.
Azerbaijan and OpenAI Agree to Create an AI Platform for Schools
The Ministry of Science and Education of Azerbaijan, the Ministry of Digital Development and Transport, and OpenAI have signed a memorandum to create an adaptive educational platform based on AI. The platform will cover more than 500,000 students and will be implemented as part of the "Digital School" project as an additional tool to the existing national education program.
Alphabet Plans to Raise $80 Billion for Artificial Intelligence Infrastructure Development
Alphabet announced plans to raise $80 billion through stock issuance. The funds will be directed toward expanding global computing power and developing AI infrastructure. The company explained the decision by stating that demand for AI solutions from businesses and users exceeds the current capabilities of the infrastructure.
Microsoft Introduces Frontier Tuning — Technology That Trains AI to Work Like You Do
At Microsoft Build 2026, the company announced Frontier Tuning, a technology for adapting AI to the specific processes of an organization. The system learns from real-world workflows, corporate data, and company regulations without affecting production systems. All configured models remain strictly within the organization's security perimeter. Frontier Tuning is already available in closed preview and will soon appear in Microsoft Copilot Studio and Microsoft Foundry.
Airbnb CEO Brian Chesky Plans to Open His Own AI Laboratory
Airbnb CEO Brian Chesky is founding a new artificial intelligence laboratory. The likely focus will be on user interaction and design—areas Chesky has paid special attention to at Airbnb. He will not lead the laboratory directly, retaining his position as CEO of the company.
Chesky has long criticized existing AI products as not mature enough for travel and e-commerce—which is why Airbnb has not yet partnered with any large language model developers.
Azercell and AWS Developed a Platform for Training AI Models in the Azerbaijani Language
Azercell Telecom, in collaboration with Amazon Web Services, has created a platform for training large language models in the Azerbaijani language. The project is hosted on the official AWS website as a showcase example of adapting AI solutions for languages with limited digital data volume.
The platform improves training efficiency and allows working with a significantly larger corpus of Azerbaijani texts.
Google Launches Gemini Avatar for Creating a User's Digital Twin
Google has launched the Avatar feature in the Gemini app for paid subscribers. The tool creates a digital copy of the user based on their appearance and voice—setup requires looking into the camera and speaking a few numbers. The Omni model powers the feature. The avatar can be invoked directly in chats via the @me command. Each created video contains an invisible SynthID watermark. The feature is available only to users over 18 years old.
ChatGPT Receives Major Update and Will Become a "Super App" Ahead of OpenAI's IPO
The service is planned to be transformed into a "super app" combining programming tools and AI agents.
This move is linked to preparations for an IPO, which could take place in the second half of 2026—the company is strengthening its work with large corporate clients amid competition from Anthropic. Initial changes are expected in the coming weeks. The IPO timeline has not yet been officially disclosed.
The Tech on Telegram Follow The Tech on Telegram
Keep up with the latest technology and startup news.