Open to anyone with an idea
Microsoft for Startups Founders Hub brings people, knowledge, and benefits together to help founders at all stages solve their startup challenges. Sign up in minutes, no funds required.
This week at Microsoft Ignite, we showcased the latest advancements in Microsoft AI and cloud technologies. Over the course of two days, Microsoft announced more than 100 updates covering generative AI infrastructure innovations, new tools, adoption, productivity, and security improvements. These updates are game-changing for everyone building AI and demonstrate Microsoft Azure’s commitment to becoming the leading cloud platform for AI.
We’re so excited about the innovation these updates will bring to the ecosystem, so we’ve rounded up the top five Microsoft AI announcements every startup should know about.
- Microsoft AI Supercomputer: Get more power, better prices, and new purpose-built silicon through our Nvidia partnership.
- Simplifying OSS LLM interactions: Simplified interaction through easy API calls (Llama2, Mistral, Jais)
- Azure OpenAI service updates: Enhance Azure OpenAI services with multi-mode capabilities, OpenAI GPT-4 fine-tuning, OpenAI GPT-4 Turbo with Vision, and cost-effective pricing.
- New vector search and semantic ranking: Prioritizing accuracy in information retrieval
- Enhanced Responsible AI Tools: We promote the development of applications that build consumer and business trust through advanced security measures and ethical considerations.
Azure AI Infrastructure – Azure is the world’s computer.
Azure supports a variety of solutions, from cloud services to running the most sophisticated AI models. We’re excited to announce that we’ve added pricing and performance choices across the Azure infrastructure technology stack in your data centers and the silicon that supports them, including:
- Custom silicon for AI Enterprise workloads on Microsoft Cloud. This new custom silicon complements the products Microsoft offers with industry partners. The two new chips, Microsoft Azure Maia and Microsoft Azure Cobalt, are built with a holistic view of hardware and software systems to optimize performance and price. Microsoft Azure Maia is an AI Accelerator chip designed to run cloud-based training and inference for AI workloads such as OpenAI models, Bing, GitHub Copilot, and ChatGPT. Microsoft Azure Cobalt is a cloud-native chip based on Arm architecture optimized for performance, power efficiency, and cost-effectiveness for general-purpose workloads.
- Azure Boost is now generally available. Azure Boost, one of Microsoft Azure’s latest and most significant infrastructure improvements, is now generally available. Azure Boost improves large-scale network and storage performance, improves security, and improves service impact by moving virtualization processes traditionally performed on host servers, such as networking, storage, and host management, to purpose-built hardware and software optimized for these processes. Reduce . These innovations enable Microsoft to achieve the fastest remote and local storage performance on the market today, with remote storage performance of 12.5 gigabits per second (Gbps) throughput, 650K input/output operations per second (IOPS), and local storage performance of 17.3. there is. Gbps throughput and 3.8M IOPS.
- New VM and GPU products. The new GPU products include the ND MI300 v5 virtual machine with AMD chips optimized for generative AI workloads, the NC H100 v5 virtual machine with the latest NVIDIA GPUs, and the new NC H100 v5 virtual machine (VM) built on the latest NVL. ) includes a preview of the series. A variant of the NVIDIA Hopper 100 (H100) that offers more memory per GPU. These new VM series will provide customers with greater performance, reliability, and efficiency for mid-level AI training and generative AI inference. By maintaining more memory per GPU in VMs, customers increase data processing efficiency and improve overall workload performance.
Introduction to service model
Azure Machine Learning continues to improve the user experience with new enhancements, including general availability of prompt flows and model catalogs and preview of OneLake integration with Microsoft Fabric, helping startups and machine learning experts simplify AI-based application development and create AI. Operationalize responsible generative AI solutions at every stage of the development lifecycle. One of the key new features is the Model-as-a-Service feature in the Model Catalog, which allows startups to easily integrate the latest AI models, such as Meta’s Llama 2, the upcoming premium model from Mistral, and G42’s Jais, into the API. no see. This is the endpoint for your application. Startups can also customize models with their own data without having to worry about setting up and managing GPU infrastructure, helping remove complexity and increase productivity, which is critical for early-stage startups working with small teams.
New multi-model capabilities have been added to the Azure OpenAI service.
This has been one of the most requested areas for startups on Founders Hub over the past few weeks, and we’re excited to see this feature coming to Azure. The new OpenAI GPT-3.5 Turbo model with 16K token prompt length will be generally available, and OpenAI GPT-4 Turbo will be available in public preview on the Azure OpenAI service in late November 2023. OpenAI GPT-4 Turbo allows customers to shorten the length and provide more control and efficiency to their generative AI applications. Additionally, OpenAI GPT-4 Turbo with Vision will soon be released in preview, and DALL·E 3 is now available in public preview on the Azure OpenAI service. Together with OpenAI GPT-4, this can fuel next-generation enterprise solutions, allowing startups to pursue advanced technologies. Features with images. Additionally, when used with the Azure AI Vision service, OpenAI GPT-4 Turbo with Vision also understands video to generate text output.
General availability of vector search and semantic ranker
Azure AI Search (formerly Azure Cognitive Search) is an information discovery and discovery platform that enables startups to deliver highly personalized experiences in generative AI applications. Vector Search, a feature of Azure AI Search, is now generally available, enabling startups to create highly accurate experiences for all users in their generative AI applications. Semantic Ranker (formally called Semantic Search), which includes a multilingual deep learning model adapted from Microsoft Bing, is also available, allowing startups to prioritize the most relevant search results and ensure they are served first.
Strengthening our commitment to responsible AI
Finally, all startups building generative AI solutions face the new challenge of building their products in a safe and responsible way. Microsoft is leading the industry in the safe and responsible use of AI. The company sets the standard with its Copilot Copyright Commitment (CCC), an industry-leading commitment to protect and indemnify commercial customers from copyright infringement lawsuits. Today, Microsoft takes that commitment one step further by announcing the expansion of CCC to customers using the Azure OpenAI service. The new benefit is called the Customer Copyright Agreement. Additionally, Azure AI Content Safety is now generally available, helping startups further detect and mitigate harmful content and create better online experiences. Customers can use Azure AI Content Safety as a built-in safety system within the Azure OpenAI service, as part of prompted engineering in Azure Machine Learning for open source models, or as a standalone API service.
Startups can access Azure AI for free through Microsoft for Startups.
To learn more about these services and ~95 other announcements, Microsoft Ignite is available on-demand here. Founders can take advantage of up to $150,000 in credits toward Azure AI services through the Microsoft for Startups Founders Hub program to test and explore these capabilities for themselves. Additionally, our AI advisory team helps startups accelerate their AI journey through one-on-one consultations.
We’re excited to introduce these innovative features in Microsoft Ignite 2023 and look forward to the exciting possibilities they bring.
Not yet signed up for Microsoft for Startups Founders Hub? Sign up now.