Welcome to NexaQuanta’s Latest AI Innovations Newsletter!
This edition brings you some of the most exciting developments shaping the AI landscape, from groundbreaking collaborations to significant cost reductions in AI models. Here’s a sneak peek at what’s inside:
1). IBM Unveils DNS Performance Insights: Discover how IBM and Catchpoint Systems are analysing the performance of DNS providers during peak traffic times, revealing critical differences that could impact your digital infrastructure.
2). Granite Code Models on NVIDIA NIM Microservices: IBM’s powerful Granite code models are now available as NVIDIA-hosted microservices, offering scalable AI solutions for enterprise code generation. Find out how this integration is revolutionising generative AI for businesses.
3). OpenAI Slashes GPT-4 Token Prices: Learn about OpenAI’s dramatic reduction in GPT-4 token costs, which is set to reshape the competitive landscape in the AI market and drive further innovation.
4). Intel and IBM’s Strategic AI Collaboration: Read about the new partnership between IBM and Intel to deploy Gaudi 3 AI accelerators on IBM Cloud, paving the way for more cost-effective and secure enterprise AI solutions.
IBM Unveils Key Findings on DNS Performance in Collaboration with Catchpoint Systems
In partnership with Catchpoint Systems, Inc., IBM has released new insights into the performance of DNS providers during peak internet traffic periods. Through a series of tests designed to replicate real-world conditions, the study reveals significant performance differences among DNS providers, particularly highlighting the slower speeds of self-hosted DNS architectures compared to managed services.
The report underscores the critical role of DNS performance in maintaining fast website load times. IBM notes that 1 in 4 visitors will abandon a site that takes more than 4 seconds to load. The collaboration between IBM and Catchpoint provides a detailed analysis of global DNS connections, offering valuable data for businesses aiming to optimise their digital infrastructure.
Check details here.
IBM Granite Code Models Now Available as NVIDIA NIM Microservices
IBM has announced the availability of its Granite code models as NVIDIA-hosted NIM inference microservices, marking a significant step in scaling enterprise generative AI (gen AI) for code generation. This development was unveiled at the NVIDIA AI Summit in Taiwan, where IBM highlighted the role of these models in modernising enterprise code and delivering measurable ROI.
The IBM Granite family, known for its multilingual language and code capabilities, includes the 8b-code-instruct and 34b-code-instruct models optimised for high performance and throughput with NVIDIA NIM. These models are now accessible via the NVIDIA API catalogue, offering enterprises a powerful tool for deploying AI models across GPU-accelerated infrastructures.
IBM’s collaboration with NVIDIA aims to accelerate gen AI adoption, combining NVIDIA AI Enterprise software with IBM’s industry solutions. Granite code models have been recognized for their transparency and performance, outperforming larger models in evaluations such as the HumanEvalPack. They support 116 programming languages and are available on platforms like Hugging Face, GitHub, and watsonx.ai.
With the new availability of the NVIDIA API catalogue, IBM is empowering developers to deploy trusted, cost-effective AI solutions at scale, ensuring data privacy and security through flexible deployment options. As IBM and NVIDIA continue to strengthen their partnership, more Granite models are expected to be added.
OpenAI Slashes GPT-4 Token Prices, Sparking Competitive Pressure in AI Market
OpenAI has announced a significant reduction in the cost of GPT-4 tokens. They are now priced at $4 per million tokens, down from $36 at their initial launch in March 2023. This 79% annual price drop reflects the rapidly changing landscape of generative AI, driven by advances in both software and hardware.
The reduced pricing is further impacted by the rise of open-weight models like Llama 3.1, which allow API providers, including startups like Anyscale and Together.ai, to compete aggressively on price without the burden of recouping model development costs. This competitive environment, combined with hardware innovations from companies such as Groq and NVIDIA, is expected to lower token prices.
This trend signals a shift in focus for AI developers and companies. Rather than optimising costs, the priority should be building robust, useful applications to take advantage of falling token prices over time. As new models emerge and token prices continue to decline, even currently marginally expensive applications may soon become economically viable.
Moreover, switching between models, particularly those with open weights, could further enhance cost efficiency. However, challenges remain, particularly in implementing effective regression testing when transitioning to new models. As evaluation methods improve, these barriers will likely diminish, making it easier for companies to adapt to the evolving AI landscape.
Open the link for more details.
Intel and IBM Collaborate to Provide Better Cost Performance for AI Innovation
IBM and Intel have announced a strategic collaboration to deploy Intel® Gaudi® 3 AI accelerators as a service on IBM Cloud, with availability expected in early 2025. This partnership aims to make enterprise AI more cost-effective and drive innovation while ensuring security and resiliency. IBM Cloud will be the first cloud service provider to adopt Gaudi 3, offering accelerators for hybrid and on-premise environments.
Integrating Gaudi 3 AI Accelerators with IBM’s watsonx AI and data platform will enhance enterprise AI workloads’ support. This move is designed to help businesses scale AI capabilities more affordably and efficiently, addressing the growing demands for high performance, cost efficiency, and security in AI computing.
Intel’s executive vice president and general manager, Justin Hotard, emphasised the importance of an open and collaborative ecosystem in unlocking AI’s full potential. He noted that integrating Gaudi 3 AI Accelerators and Xeon CPUs with IBM Cloud would create new AI capabilities and meet the increasing demand for innovative and secure AI solutions.
For IBM, this collaboration aligns with its commitment to driving AI and hybrid cloud innovation. Alan Peacock, GM of IBM Cloud, highlighted that using Intel’s Gaudi 3 accelerators will offer clients a flexible AI solution optimised for cost performance, enabling new AI business opportunities.
With this partnership, IBM and Intel are set to empower enterprises to scale AI workloads more effectively. It will also optimise inferencing costs and innovate within secure, resilient cloud environments.
Read further details here.
Thank You for Staying Ahead with NexaQuanta!
As we wrap up this edition of our AI Innovations Newsletter, we hope the insights shared have sparked new ideas and possibilities for your business. At NexaQuanta, we’re committed to informing you about the latest trends and advancements that can drive your success.
Don’t miss out on future updates, expert analysis, and exclusive content to help you stay competitive. Subscribe to our newsletter today and join a community of forward-thinking professionals dedicated to harnessing the power of AI. Together, we can explore new frontiers and unlock the full potential of technology for your enterprise.