
NVIDIA's Granary Dataset & Rigetti's Quantum Leap: AI's Next Frontier - Silicon Brief #31
August 18, 2025 -
Silicon Brief
Hi there,
In this edition of Silicon Brief, we delve into the latest developments in AI chips and data centers. NVIDIA has released the Granary Dataset and new speech AI models, marking a significant step in AI technology advancements. Meanwhile, Rigetti Computing has launched a 36-qubit quantum computer on the cloud, showcasing the growing intersection of quantum computing and AI. These innovations highlight the rapid pace of technological progress in the AI hardware sector.
On the data center front, Rio has partnered with Oracle and Nvidia for an ambitious AI city project, while Vast Data and SK Telecom are collaborating on an AI cluster. Additionally, DayOne is set to build a EUR 1.2 billion data center in Finland, reflecting the increasing demand for robust AI infrastructure. These projects underscore the critical role of data centers in supporting the burgeoning AI industry.
In this edition of Silicon Brief, we delve into the latest developments in AI chips and data centers. NVIDIA has released the Granary Dataset and new speech AI models, marking a significant step in AI technology advancements. Meanwhile, Rigetti Computing has launched a 36-qubit quantum computer on the cloud, showcasing the growing intersection of quantum computing and AI. These innovations highlight the rapid pace of technological progress in the AI hardware sector.
On the data center front, Rio has partnered with Oracle and Nvidia for an ambitious AI city project, while Vast Data and SK Telecom are collaborating on an AI cluster. Additionally, DayOne is set to build a EUR 1.2 billion data center in Finland, reflecting the increasing demand for robust AI infrastructure. These projects underscore the critical role of data centers in supporting the burgeoning AI industry.
Energy & Sustainability
Rolls-Royce CEO Tufan Erginbilgic has highlighted the potential of small modular nuclear reactors to power AI, positioning the company as a leader in meeting AI's energy needs with contracts in the UK and Czech Republic. Meanwhile, a report by EPRI and Epoch AI forecasts that electricity demand for training large AI models could exceed 50 gigawatts by 2030, driven by the increasing complexity of AI models.
- Rolls-Royce CEO Discusses Nuclear-Powered AI Plans
- AI Model Training Power Demand to Exceed 50 GW by 2030
- Consumers Energy Prepares for Increased Power Demand from AI-Driven Data Centers
- EDP Leads GREEN.DAT.AI Project for Energy Sector AI Efficiency
- US Department of Energy Selects 11 Nuclear Projects
- Equinix Collaborates with Energy Firms for Sustainable Data Centers
Technology
Mistral AI has achieved a 2.5x increase in model training speeds by utilizing NVIDIA's GB200 NVL72 infrastructure through CoreWeave.
Dell Technologies has updated its AI Data Platform to improve access to unstructured data for generative AI applications, featuring a new data engine developed with Elastic and the PowerEdge R7725 server with Nvidia GPUs.
Dell Technologies has updated its AI Data Platform to improve access to unstructured data for generative AI applications, featuring a new data engine developed with Elastic and the PowerEdge R7725 server with Nvidia GPUs.
- Mistral AI Boosts Training Speeds with NVIDIA and CoreWeave
- Dell Enhances AI Platform for Generative AI
Regulation & Policy
The U.S. government is considering investing in Intel Corp. to bolster domestic manufacturing, focusing on a delayed chip factory in Ohio. Meanwhile, U.S. authorities are monitoring AI chip shipments from Dell and Supermicro to prevent diversion to China, aligning with export restrictions to limit military access to advanced technology.
- U.S. Government May Invest in Intel for Manufacturing
- US Monitors AI Chip Shipments to Prevent Diversion to China
- U.S. Government to Receive 15% from Nvidia and AMD China Sales
Data Centers
Rio de Janeiro City Hall has partnered with Oracle and Nvidia to launch the 'Rio AI City' project, aiming to position the city as a global AI hub. Meanwhile, Vast Data and SK Telecom are collaborating on the Haein AI cluster in South Korea, utilizing Nvidia Blackwell GPUs to offer GPU-as-a-Service, enhancing AI infrastructure across various sectors.
- SK Telecom to Develop AI Infrastructure with Haein Cluster
- DayOne to Invest EUR 1.2 Billion in Finnish Data Centre
- NVIDIA and NSF Collaborate with Ai2 on AI Models for Science
- ODATA Opens Fourth Hyperscale Data Center in Mexico
- CoreWeave Reports $1.21 Billion Revenue Amid AI Demand but Faces $290.5 Million Loss
- Rumble Evaluates $1.17 Billion Bid for Northern Data
- Rio de Janeiro Partners with Oracle and Nvidia for AI Project
- Vast Data and SK Telecom to Develop AI Cluster in South Korea
- Johor Faces Resource Strain from AI Data Centers
- Vesper Fund Acquires Majority Stake in Terakraft AS
- Private Equity Investment in AI Infrastructure
- DC Blox Secures $1.15 Billion for Georgia Data Center
- LG CNS to Build AI Data Center in Vietnam
- MARA Acquires Majority Stake in EDF's Exaion
- G42 Launches OpenAI Models on Core42 AI Cloud
- Edgemode and Vertical Data Plan AI Data Center in Sweden
- CMC Corporation and Samsung C&T Sign $1 Billion Data Center Deal
- Blackstone's QTS Secures $1.65 Billion for Data Centers
- Google Invests $9 Billion in Oklahoma AI and Cloud Expansion
- AVAIO Digital Reaches Milestone for AI-Ready Data Hub in Madrid
- Oracle Partners with Google to Offer Gemini Models
- Digital Alpha Invests $300 Million in Massed Compute
- Sime Darby Property Seeks Loan for Google Data Center in Malaysia
- Galaxy Digital Secures $1.4 Billion for Helios AI Datacenter
Chips & Hardware
NVIDIA has released the Granary dataset, featuring 1 million hours of audio to enhance multilingual speech recognition and translation for 25 European languages. Meanwhile, Rigetti Computing has launched its 36-qubit quantum computer, Cepheus-1-36Q, on its cloud platform, with plans to offer it on Microsoft Azure soon.
- NVIDIA Releases Granary Dataset and New AI Models
- Rigetti Computing Launches 36-Qubit Quantum Computer
- NeoLogic Secures $10M for Energy-Efficient AI Processors
- EDGX Secures €2.3 Million for Satellite AI Computing
- SkyWater Technologies Acquires Infineon's Fab 25 in Texas
- Tesla Ends Dojo AI Supercomputer Project
- Syenta Secures $8.8M for AI Semiconductor Tech
- Aitech Unveils A230 Vortex AI Supercomputer for Military Use
- Huawei Unveils Unified Cache Manager Algorithm
- Normal Computing Completes Tape-Out of CN101 Chip
- Rivos Seeks Up to $500 Million to Challenge Nvidia
- HPE Expands AI Server Range with NVIDIA Blackwell GPUs
- DeepSeek Delays AI Model Release Due to Huawei Chip Issues
- Foxconn Sees 27% Profit Rise Due to AI Demand
- DeepSeek Switches to Nvidia for R2 Model
We hope you enjoyed this article.
Consider subscribing to one of our newsletters like Silicon Brief or Daily AI Brief.
Also, consider following us on social media:
More from: Data Centers
08/18
Oracle Integrates OpenAI GPT-5 Across Cloud and Database Services
08/18
Applied Digital Plans 280MW AI Data Center in North Dakota
08/16
Galaxy Secures $1.4 Billion for Helios AI Datacenter Expansion
08/14
DeepSeek Delays AI Model Due to Huawei Chip Issues
08/14
Equinix Partners with Alternative Energy Providers for AI Data Centers
More from: Data Centers
08/18
Applied Digital Plans 280MW AI Data Center in North Dakota
08/16
Galaxy Secures $1.4 Billion for Helios AI Datacenter Expansion
08/14
Equinix Partners with Alternative Energy Providers for AI Data Centers
08/14
AVAIO Digital Advances AI-Ready Data Center in Madrid
08/13
EdgeMode and Vertical Data to Develop AI-Optimized Data Center in Sweden