News from the AI & ML world
DeeperML - #aiinfrastructure
|
Allyson Vasquez@NVIDIA Technical Blog
//
NVIDIA's GTC 2025 is shaping up to be a major event for AI enthusiasts, packed with networking opportunities, live demos, and discussions on the latest AI innovations. Data Phoenix is highlighting the event as a key gathering, featuring meetups, networking receptions, and hands-on sessions alongside the main conference. They are also co-hosting and supporting key events like the INFRA@GTC Networking Reception and AI Demo Jam.
VAST Data plans to showcase its data platform for enterprise Retrieval Augmented Generation (RAG) use cases at the conference. Microsoft and NVIDIA have also announced a partnership to integrate RTX Neural Shaders into a DirectX preview in April, bringing more AI capabilities to game development. This integration will allow developers to leverage Tensor cores in RTX GPUs to accelerate neural networks within a game's graphics pipeline. Recommended read:
References :
Jaime Hampton@BigDATAwire
//
NVIDIA's GTC 2025 showcased significant advancements in AI, marked by the unveiling of the Blackwell Ultra GPU and the Vera Rubin roadmap extending through 2027. CEO Jensen Huang emphasized a 40x AI performance leap with the Blackwell platform compared to its predecessor, Hopper, highlighting its crucial role in inference workloads. The conference also introduced open-source ‘Dynamo’ software and advancements in humanoid robotics, demonstrating NVIDIA’s commitment to pushing AI boundaries.
The Blackwell platform is now in full production, meeting incredible customer demand, and the Vera Rubin roadmap details the next generation of superchips expected in 2026. Huang also touted new DGX systems, highlighting the push towards photonic switches to handle growing data demands efficiently. Blackwell Ultra will offer 288GB of memory. NVIDIA claims the GB300 chip brings 1.5x more AI performance than the NVIDIA GB200. These advancements aim to bolster AI reasoning capabilities and energy efficiency, positioning NVIDIA to maintain its dominance in AI infrastructure. Recommended read:
References :
@tomshardware.com
//
Nvidia has unveiled its next-generation data center GPU, the Blackwell Ultra, at its GTC event in San Jose. Expanding on the Blackwell architecture, the Blackwell Ultra GPU will be integrated into the DGX GB300 and DGX B300 systems. The DGX GB300 system, designed with a rack-scale, liquid-cooled architecture, is powered by the Grace Blackwell Ultra Superchip, combining 36 NVIDIA Grace CPUs and 72 NVIDIA Blackwell Ultra GPUs. Nvidia officially revealed its Blackwell Ultra B300 data center GPU, which packs up to 288GB of HBM3e memory and offers 1.5X the compute potential of the existing B200 solution.
The Blackwell Ultra GPU promises a 70x speedup in AI inference and reasoning compared to the previous Hopper-based generation. This improvement is achieved through hardware and networking advancements in the DGX GB300 system. Blackwell Ultra is designed to meet the demand for test-time scaling inference with a 1.5X increase in the FP4 compute. Nvidia's CEO, Jensen Huang, suggests that the new Blackwell chips render the previous generation obsolete, emphasizing the significant leap forward in AI infrastructure. Recommended read:
References :
Maximilian Schreiner@THE DECODER
//
OpenAI has announced it will adopt Anthropic's Model Context Protocol (MCP) across its product line. This surprising move involves integrating MCP support into the Agents SDK immediately, followed by the ChatGPT desktop app and Responses API. MCP is an open standard introduced last November by Anthropic, designed to enable developers to build secure, two-way connections between their data sources and AI-powered tools. This collaboration between rivals marks a significant shift in the AI landscape, as competitors typically develop proprietary systems.
MCP aims to standardize how AI assistants access, query, and interact with business tools and repositories in real-time, overcoming the limitation of AI being isolated from systems where work happens. It allows AI models like ChatGPT to connect directly to the systems where data lives, eliminating the need for custom integrations for each data source. Other companies, including Block, Apollo, Replit, Codeium, and Sourcegraph, have already added MCP support, and Anthropic's Chief Product Officer Mike Krieger welcomes OpenAI's adoption, highlighting MCP as a thriving open standard with growing integrations. Recommended read:
References :
Chris McKay@Maginative
//
NVIDIA's GTC 2025 event showcased significant advancements in AI infrastructure, highlighting the Blackwell Ultra and Rubin architectures, along with several related technologies and partnerships. Jensen Huang, Nvidia CEO, delivered a keynote address outlining the company’s vision for the AI-powered future, emphasizing improvements in processor performance, network design, and memory capabilities. The Blackwell Ultra GPUs are being integrated into DGX systems to meet the rising demands of AI workloads, especially in inference and reasoning.
NVIDIA is also expanding its offerings beyond chips with the introduction of desktop AI supercomputers for developers. The DGX Station, powered by the GB300 Blackwell Ultra Superchip, aims to bring data center-level AI capabilities to a compact form factor. Nvidia introduced Dynamo, an open-source inference software engineered to maximize token revenue generation for AI factories deploying reasoning AI models. The presentation emphasized a clear roadmap for data center computing, advancements in AI reasoning capabilities, and bold moves into robotics and autonomous vehicles. Recommended read:
References :
Matt Milano@WebProNews
//
OpenAI has inked a significant five-year, $11.9 billion agreement with CoreWeave, a cloud infrastructure provider specializing in AI workloads. This substantial investment aims to secure the necessary GPU compute capacity for OpenAI's increasingly demanding AI models. The deal also includes OpenAI acquiring a $350 million stake in CoreWeave, demonstrating a deeper strategic partnership between the two companies as CoreWeave prepares for its IPO.
This move signifies OpenAI's ongoing quest for ever greater AI compute capabilities and a diversification of its cloud infrastructure strategy beyond Microsoft Azure. CoreWeave, backed by NVIDIA, operates 32 AI data centers housing over 250,000 GPUs and is positioning itself as a key player in the AI infrastructure space. The partnership also strengthens CoreWeave’s market position as it attempts to reduce its reliance on Microsoft, which accounted for 62% of its $1.92 billion revenue in 2024. Recommended read:
References :
staff@insideAI News
//
OpenAI and Oracle are collaborating on a massive AI data center project in Texas, known as 'Stargate', which will be equipped with Nvidia's GB200 Blackwell chips. The data center, located in Abilene, Texas, aims to become a transformative benchmark for next-generation AI innovation.
This ambitious project involves the purchase of 64,000 Nvidia GB200 chips, to be installed in phases with an initial rollout of 16,000 chips planned for completion within the next six months. The data center is expected to house all 64,000 of the chips by the end of 2026. The GB200 Superchip, which combines a Grace CPU with two enhanced B200 GPUs and priced between $60,000-$70,000 per chip, will provide the necessary computing power for the facility. An OpenAI spokesperson confirmed their collaboration with Oracle on the design and delivery of the data center, noting that Oracle will oversee the acquisition and operation of the supercomputer being constructed. This venture is part of a larger $100 billion Stargate infrastructure initiative. Recommended read:
References :
staff@insideAI News
//
Celestial AI, the creator of the Photonic Fabric optical interconnect technology, has announced a successful Series C1 funding round, securing $250 million. The round was led by Fidelity Management & Research Company, bringing the total capital raised by the company to over $515 million. Celestial AI's Photonic Fabric platform aims to speed up data transfer inside servers using light, with plans for new data centers in North America and France. This investment underscores the growing demand for advanced AI models and the development of AI infrastructure.
The company's Photonic Fabric technology allows AI compute to be networked seamlessly, from within processor packages to servers across multiple racks. Celestial AI offers a full suite of products, that include connectivity, switching and packaging solutions which serve as the foundation for optical scale-up networks for accelerated computing. Celestial AI's valuation has now reached $2.5 billion, reflecting investor confidence in its potential to revolutionize AI infrastructure. New investors include funds and accounts managed by BlackRock, Maverick Silicon, Tiger Global Management and Lip-Bu Tan, as well as participation from existing investors including AMD Ventures, Koch Disruptive Technologies (KDT), Temasek, Temasek’s wholly-owned subsidiary Xora Innovation, Porsche Automobil Holding SE and The Engine Ventures. Recommended read:
References :
Harsh Mishra@Analytics Vidhya
//
DeepSeek AI has been making significant contributions to the open-source community, particularly in the realm of AI model efficiency and accessibility. They recently launched the Fire-Flyer File System (3FS), a high-performance distributed file system tailored for AI training and inference workloads. This system is designed to address the challenges of managing large-scale, concurrent data access, a common bottleneck in traditional file systems. 3FS leverages modern SSDs and RDMA networks, offering a shared storage layer that facilitates the development of distributed applications by bypassing limitations seen in more traditional, locality-dependent file systems.
DeepSeek's commitment extends to data processing and model optimization. They have introduced the Smallpond framework for data processing and released quantized DeepSeek-R1 models, optimized for deployment-ready reasoning tasks. The quantized models, including Llama-8B, Llama-70B, Qwen-1.5B, Qwen-7B, Qwen-14B, and Qwen-32B, are available as a Hugging Face collection with evaluations, benchmarks, and setup instructions. These models maintain competitive reasoning accuracy while unlocking significant inference speedups. Recommended read:
References :
Jaime Hampton@AIwire
//
Cerebras Systems is significantly expanding its AI infrastructure to challenge Nvidia's dominance in the AI market. The company is deploying over a thousand of its wafer-scale AI accelerator chips across six new data centers in North America and France. This expansion aims to provide ultrafast AI inference capabilities, promising faster speeds and cost reductions compared to traditional GPU-based setups.
These new data centers will process an impressive 40 million tokens per second, with 85% of the capacity located in the United States. Facilities are already operational in Santa Clara, Stockton, and Dallas, and further expansion includes sites in Minneapolis (Q2 2025), Oklahoma City and Montreal (Q3), and Atlanta and France (Q4). Cerebras is also partnering with Hugging Face to provide developers with easy access to its AI inference service, marking a major distribution channel for open-source models like Llama 3. Recommended read:
References :
Ellie Ramirez-Camara@Data Phoenix
//
Nvidia is making significant strides in the realm of AI agents, highlighted at this year's GTC 2025 conference. CEO Jensen Huang emphasized the transformative impact of agentic AI and reasoning models, predicting that these technologies will revolutionize industries and automate processes. To support this shift, Nvidia unveiled the Blackwell Ultra platform, designed to handle the demanding requirements of AI reasoning, agentic AI, and physical AI applications. The platform, which includes the GB300 NVL72 rack-scale solution and the HGX B300 NVL16 system, offers substantial performance improvements over previous generations, with the GB300 NVL72 delivering 1.5x more AI performance.
In addition to hardware advancements, Nvidia launched NVIDIA Dynamo, an open-source inference framework, to optimize reasoning AI services across thousands of GPUs. This framework is designed to maximize token revenue generation for AI factories deploying reasoning AI models by orchestrating and accelerating inference communication across GPU clusters. Major cloud providers and server manufacturers are expected to offer Blackwell Ultra-based products starting in the second half of 2025. These developments position Nvidia as a key player in the emerging landscape of AI agents and reasoning models, promising to drive significant advancements in AI capabilities and applications. Recommended read:
References :
@openai.com
//
References:
GZERO Media
, www.marketingaiinstitute.com
OpenAI has recently partnered with the US National Laboratories to provide its AI models for applications in national security and scientific research. This move aims to leverage the power of artificial intelligence in critical areas, enhancing capabilities in both research and security sectors. The collaboration underscores the growing recognition of AI's potential to address complex challenges and drive innovation across various domains.
France is making significant investments to strengthen its position as an AI hub. President Emmanuel Macron announced that foreign and local companies will invest €109 billion in AI projects within the country. This financial commitment includes €20 billion from Brookfield, with additional financing from the UAE potentially reaching €50 billion. California State University just made a massive move in higher ed that might set the tone for how colleges nationwide adopt AI. The 23-campus system, serving more than 460,000 students and 63,000 staff and faculty, is rolling out a specialized version of ChatGPT—called ChatGPT Edu—to all of them. Recommended read:
References :
@www.cnbc.com
//
References:
africa.businessinsider.com
, techcrunch.com
,
Meta is significantly increasing its investment in artificial intelligence, with CEO Mark Zuckerberg pledging "hundreds of billions" of dollars in long-term spending. This strategic move comes as Meta reports a strong fourth quarter, boasting a 21% year-over-year revenue increase to $48.4 billion and a 49% jump in net income to $20.8 billion. Zuckerberg views this massive investment in AI infrastructure as a crucial "strategic advantage" for Meta's future, enabling them to compete effectively and serve their billions of users. This move is in part a response to the emergence of new competitors like DeepSeek.
Meta's Reality Labs, while still operating at a loss of $4.97 billion in Q4, has shown positive signs with revenue up 1% year-over-year to $1.1 billion. Furthermore, internal memos reveal that Reality Labs surpassed nearly all sales and user targets for 2024, experiencing a 40% overall sales growth. Meta is particularly focused on developing open-source AI models, aiming to make Llama 4 the most competitive in the world. This open-source strategy is seen as a way to allow Meta to innovate and compete with established AI leaders, despite recent market anxieties regarding DeepSeek. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |