Google has announced a new cloud platform that allows users to access the most advanced Gemini models on their devices securely. Called Private AI Compute, the platform works similar to Apple’s Private Cloud Compute, acting like “one seamless Google stack,” powered by the Tensor Processing Units or TPUs. 

The platform processes the same type of sensitive data that is expected during on-device processing. However, since AI tools need more compute, it isn’t possible to process on-device. 

The TPUs depend on an AMD-based Trusted Execution Environment (TEE) which encrypts and isolates the memory from even Google. 

In a blog posted by the company, Google also said that Private AI Compute helps extended capabilities as well like Magic Cue, a tool which offers timely suggestions on the latest Pixel 10 phones. 

Additionally, the Recorder app on the Pixel will also be able to summarise transcriptions across multiple languages.

Advanced Micro Devices said on Tuesday that it expected annual data center chip revenue of $100 billion within the next five years, and its earnings to more than triple.

The Santa Clara, California-based chip designer’s shares were up 4% in choppy post-market trading, after closing down 2.7% at $237.52. The stock has risen 16% since October 6, when the company signed a lucrative multiyear deal with OpenAI that would bring in tens of billions of dollars in annual revenue.

While the deal is unlikely to dent Nvidia’s dominance in AI chipmaking, it is seen as a big vote of confidence in AMD’s chips, and the company’s bullish financial projections on Tuesday should help assuage investor concern over AMD’s ability to claw away business.

AMD expects the market for the company’s data center chips to grow to $1 trillion by 2030, CEO Lisa Su said at its analyst day – its first such event in three years – in New York. Artificial intelligence will drive much of the growth to the trillion-dollar figure. That market includes AMD’s plain processor and networking chips, along with its specialized AI chips, Su said.

“It’s an exciting market,” Su said. “There’s no question, data center is the largest growth opportunity out there, and one that AMD is very, very well positioned for.”

In the next three to five years, AMD expects 35% growth across its entire business each year and 60% in its data center business, finance chief Jean Hu said at the analyst day.

The company also expects earnings to rise to $20 a share in the same three-to-five-year period. LSEG estimates peg AMD’s 2025 profit at $2.68 per share.

Jensen Huang, CEO of AMD archrival Nvidia, has said the broader AI infrastructure market will grow to $3 trillion to $4 trillion by 2030.

MORE SMALL M&A EXPECTED

AMD’s next-generation MI400 series of AI chips is set to launch in 2026 and include several variants designed for scientific applications and for generative AI. Along with the MI400 chips, AMD is also planning to launch a complete server rack, similar to a product Nvidia sells called the GB200 NVL72.

In her opening remarks Su highlighted the company’s recent AI-related acquisitions, including the server builder ZT Systems and a slew of smaller software companies. AMD has built “an M&A machine,” Su said.

In recent months, AMD has acquired a batch of startups that focus on building software needed to run AI applications. On Monday, AMD said it bought MK1. The plan is to ensure AMD has access to the appropriate software and the people it needs to build its AI capabilities, Chief Strategy Officer Mat Hein told Reuters in an interview.

“We’ll continue to do AI software tuck-ins,” said Hein. The chip designer forecast fourth-quarter revenue that topped Wall Street estimates. Demand for AI chips gave AMD executives a reason for optimism about the remainder of the year. The company’s data center CPU business has also benefitted from the surge in AI-related spending.

Alphabet’s Google said on Tuesday that it will invest 5.5 billion euros ($6.41 billion) in Germany in the coming years in a push to expand its infrastructure and data centre capacity in Europe’s largest economy.

The plans include a new data centre in Dietzenbach, close to Frankfurt, Google said, according to information it provided ahead of a press conference in Berlin.

OpenAI’s chatbot ChatGPT violated German copyright laws by reproducing lyrics from songs by best-selling musician Herbert Groenemeyer and others, a court ruled on Tuesday, in a closely watched case against the U.S. firm over its use of lyrics to train its language models.

The regional court in Munich found that the company trained its AI on protected content from nine German songs, including Groenemeyer’s hits “Maenner” and “Bochum”.

The case was brought by German music rights society GEMA, whose members include composers, lyricists and publishers, in another sign of artists around the world fighting back against data scraping by AI.

Presiding judge Elke Schwager ordered OpenAI to pay damages for the use of copyrighted material, without disclosing a figure.

GEMA legal advisor Kai Welp said GEMA hoped discussions could now take place with OpenAI on how copyright holders can be remunerated.

COPYRIGHT INFRINGED

OpenAI had argued that its language models did not store or copy specific training data but, rather, reflected what they had learned based on the entire training data set.

Since the output would only be generated as a result of user inputs known as prompts, it was not the defendants, but the respective user who would be liable for it, OpenAI had argued.

However, the court found that both the memorisation in the language models and the reproduction of the song lyrics in the chatbot’s outputs constitute infringements of copyright exploitation rights, according to a statement on the ruling.

POTENTIAL PRECEDENT

The outcome of the case could set a precedent in Europe for how AI companies use copyrighted materials.

“The internet is not a self-service store, and human creative achievements are not free templates,” said GEMA CEO Tobias Holzmueller. “Today, we have set a precedent that protects and clarifies the rights of authors: even operators of AI tools such as ChatGPT must comply with copyright law.”

The decision can be appealed.

“We disagree with the ruling and are considering next steps,” a spokesperson for OpenAI said. “The decision is for a limited set of lyrics and does not impact the millions of people, businesses and developers in Germany that use our technology every day.”

Earlier this year, leading Bollywood music labels asked a New Delhi court to join a copyright lawsuit against OpenAI over alleged unauthorised use of sound recordings to train AI models, underscoring global concerns about AI and music rights.

Google Cloud and Google DeepMind have announced a partnership with IIT Madras to support the launch of Indic Arena. The platform, meant to benchmark and rank AI models on tasks around Indian languages, is run by AI4Bharat, a research lab within IIT Madras.

Google Cloud will be providing cloud credits to help power the platform. 

“At AI4Bharat, our mission is to build AI for India’s specific needs. A critical part of this is having a neutral, standardized benchmark to understand how models are performing across our many languages,” said Mitesh Khapra, associate professor, IIT Madras.

The move is under Google Cloud’s larger plan to expand local hardware capacity for customers in India. Powered by Google’s AI Hypercomputer architecture with the latest Trillium TPUs, the goal is to help more Indian businesses and public sector organisations train and offer Gemini’s most advanced AI models here. 

Google Cloud also said that they plan to roll out the most powerful Gemini AI models in India with full data residency support which will help run batch support for Gemini 2.5 Flash for high-volume, non-real-time AI tasks at lower costs and Document AI which is currently in preview and helps automate document processing on a large scale.

Microsoft plans to invest $10 billion in artificial intelligence infrastructure at a data centre in Portugal’s port city of Sines over the next few years, marking one of the largest AI investment projects in Europe, it said on Tuesday.

The technology giant will work in partnership with developers Start Campus, AI infrastructure platform Nscale and chipmaker NVIDIA to deploy 12,600 next-generation NVIDIA Graphics Processing Units (GPUs) in Sines, 150 km (93 miles) south of Lisbon.

Start Campus, a venture between U.S. investment fund Davidson Kempner and Britain’s Pioneer Point Partners, in April announced a plan to invest 8.5 billion euros ($9.9 billion) by 2030 in a data centre hub to serve growing demand from major tech and AI companies.

One of the planned six buildings is already in operation.

“By strengthening the national AI infrastructure through collaboration with Nscale, NVIDIA, and Start Campus, we are helping to position Portugal as a benchmark for the responsible and scalable development of AI in Europe,” Microsoft’s Vice Chair and President Brad Smith said.

Portugal has large investment projects lined up for Sines to produce green energy to power energy-intensive data centres. Portugal’s Atlantic coastline positions it as a prime hub for subsea cables connecting Europe, Africa and the Americas and forming the backbone of the World Wide Web.

Investments in data centres, which help provide computing power for AI, have surged since OpenAI launched ChatGPT in 2022, as companies across sectors increasingly shift their operations to the cloud and integrate AI into their businesses.

Nebius Group has signed a deal worth about $3 billion with Meta to provide the Facebook owner with AI infrastructure over a five-year period, the company said on Tuesday, after it reported a more than fourfold rise in third-quarter revenue.

The company’s shares seesawed in volatile premarket trading, after it posted a surge in capital spending and a quarterly loss of over $100 million, widening from $39.7 million last year.

The stock has been on a strong run this year, with its market value rising fourfold to $27.61 billion, through last close.

The agreement with Meta underscores the surging demand for high-performance computing power that is required to build and run artificial intelligence models.

It is Nebius’ second contract with a hyperscaler, following its $17.4 billion deal with Microsoft in September.

Nebius said it would deploy the capacity needed for the Meta contract over the next three months, adding that the demand was so strong that the size of the contract had to be limited to the capacity that Nebius had available.

Amsterdam-based Nebius is among a group of so-called neocloud companies that offer hardware and cloud capacity as services. Its core business involves providing Nvidia graphics processing units and AI cloud, helping companies expand their AI infrastructure.

Nebius and its larger rival CoreWeave have seen strong demand this year as insatiable AI appetite has left even the biggest cloud companies, such as Microsoft and Amazon, with capacity constraints.

Nebius reported a 355% jump in revenue to $146.1 million in the third quarter ended September.

The company is targeting $7 billion to $9 billion in annualized run-rate revenue by the end of 2026, compared with its ARR of about $551 million at September-end.

Its capital expenditures ballooned to $955.5 million in the September quarter, from $172.1 million a year earlier, as the company invests heavily in securing GPUs, land and power.

Meta’s chief artificial intelligence scientist Yann LeCun is planning to leave the social media company to set up his own startup, the Financial Times reported on Tuesday, citing people familiar with the matter.

LeCun is also in early talks to raise funds for a new venture, according to the report.

The owner of Facebook and Instagram has significantly increased its investments in artificial intelligence, with CEO Mark Zuckerberg reorganizing the company’s AI initiatives under Superintelligence Labs.

Zuckerberg hired Alexandr Wang, former CEO of data labeling startup Scale AI to lead the new AI effort.

As a result, LeCun, who had reported to chief product officer Chris Cox, is now reporting to Wang, the report said.

Meta and LeCun did not immediately respond to Reuters requests for comment.