Latest AI news: Billions pour into AI infrastructure and collaborations

Top AI news: Asia

• Alibaba Drops Qwen 3: Open-Source Hybrid Reasoning AI Models

• Sarvam AI gets 4k GPUs to Build India’s First Homegrown GenAI Platform

• Indonesia’s Populix Raises US$4.3M Series B to Boost AI-Powered Research

• Nebius Expands AI Cloud Operations, Eyes Singapore as SEA Hub

• DeepSeek Rushes R2 Model Launch as China’s AI Race Accelerates

• Microsoft Invests $1.6bn to Supercharge Indonesia’s AI and Cloud Ambitions

Top AI news: Global

• Groq accelerates Meta Official Llama API for Developers

• Hugging Face Launches $100 Robotic Arm for DIY AI Builders

• UK Extends Qualcomm’s Bid Deadline for Alphawave

• P1 AI Raises $23M to Build Engineering Superintelligence

Stay ahead of AI with The Generative Beings

Start your AI learning and skill-building journey with a community platform having more than 1500+ member profiles, 120+ learning resources, 50+ AI tools, 50+ AI jobs, AI events, and more.

Alibaba Drops Qwen 3: Open-Source Hybrid Reasoning AI Models

Alibaba has released Qwen 3, a suite of open-source AI models with advanced hybrid reasoning, available on Hugging Face and GitHub. Qwen 3 boasts improved performance in reasoning, coding, and multilingual tasks, challenging global AI leaders.

Sarvam AI gets 4k GPUs to Build India’s First Homegrown GenAI Platform

Bengaluru-based Sarvam AI has been chosen by the Indian government, from 400 submissions, to develop the India’s first foundational large language model under the ₹10,000-crore IndiaAI Mission. Sarvam will use 4,000 high-end GPUs to build three AI models excelling in reasoning, voice, and Indian languages-entirely with local talent and infrastructure.

Indonesia’s Populix Raises US$4.3M Series B to Boost AI-Powered Research

Indonesian research startup Populix has secured US$4.3 million in Series B funding led by MSW V Asia Fund X, with participation from Intudo Ventures, Altos Ventures, and Acrew Capital. The funds will be used to advance Populix’s AI models for automating research, developing synthetic respondents for faster data collection, and expanding across Southeast Asia.

Nebius Expands AI Cloud Operations, Eyes Singapore as SEA Hub

Nebius is expanding its AI infrastructure into Southeast Asia, selecting Singapore as a key regional hub to boost AI adoption and support startups. The company will showcase its AI cloud solutions at GITEX Asia 2025, launch startup support programs, and collaborate with local enterprises, governments, and research institutions to drive regional AI growth. Backed by major investors, Nebius aims to make advanced AI resources accessible across the region.

DeepSeek Rushes R2 Model Launch as China’s AI Race Accelerates

Chinese startup DeepSeek is fast-tracking the release of its next-generation AI model, aiming to build on the momentum of its R1 model launched in January 2025. The company claims its models rival OpenAI’s in performance but are much cheaper to train-about $5.6 million versus up to $1 billion for Western counterparts.

Microsoft Invests $1.6 bn to Supercharge Indonesia’s AI and Cloud Ambitions

Microsoft has announced a $1.6 billion investment over four years to build new AI and cloud infrastructure in Indonesia, marking its largest-ever commitment in the country. The plan includes launching local data centers, training 840,000 Indonesians in AI skills, and supporting the nation’s growing developer community, all aimed at accelerating Indonesia’s digital transformation and aligning with its “Golden Indonesia 2045” vision.

Join our GenAI Demo Day in Kuala Lumpur [6 May]

AI founders and builders in Kuala Lumpur, get ready to attend our GenAI Demo Day on 6th May, along with AWS Startups. Join us for an evening filled with generative AI conversations:

• GenAI demos (ShellKode, Tripkliq, DocuAsk, 123RF & more)

• Panel discussion (Gobi Partners, Epic Angels, AWS)

• Networking (GenAI founders, builders, VCs)

Groq accelerates Meta Official Llama API for Developers

Meta and Groq have partnered to deliver the official Llama API, now accelerated by Groq’s purpose-built inference chips. This API offers first-party access to Llama 4 and other models with industry-leading speed, low latency, predictable costs, and robust privacy. Migration is simple-just three lines of code-and no prompts or data are used for training. Features include OpenAI endpoint compatibility, streaming, tool calling, and playground access.

Hugging Face Launches $100 Robotic Arm for DIY AI Builders

Hugging Face has released the SO-101, a programmable, 3D-printable robotic arm starting at $100. The SO-101 is easier to assemble than its predecessor, features improved motors, and can be trained using reinforcement learning to perform simple tasks. Prices can reach $500 for pre-assembled versions due to tariffs and supplier premiums. This launch marks Hugging Face’s push to democratize robotics and expand its open-source hardware ecosystem.

UK Extends Qualcomm’s Bid Deadline for Alphawave

The UK Takeover Panel has extended Qualcomm’s deadline to make a formal bid for British AI chipmaker Alphawave to May 12. The extension allows more time for negotiations, as Qualcomm seeks to strengthen its position in the AI semiconductor market.

P1 AI Raises $23M to Build Engineering Superintelligence

P-1 AI, co-founded by ex-Airbus CTO Paul Eremenko and ex-DeepMind researcher Aleksa Gordić, has raised $23 million led by Radical Ventures to develop “Archie,” an AI agent for automating engineering tasks in physical system design. Archie uses synthetic, physics-based datasets to master multi-physics and spatial reasoning, initially targeting data center cooling systems before expanding to industrial, automotive, and aerospace sectors.

Find the latest AI News and learning resources on TGB

Access 120+ filtered recommendations on top resources our members use to upskill on AI. These are videos, podcasts, newsletters, blogs, programs, news and more.

Join the AI founders, leaders, and builders from South-East Asia on our Telegram community channel.