Microsoft’s latest AI move will see it open up Azure AI Infrastructure for startups training LLMs

Microsoft has announced Y Combinator (YC) startups and those in its own incubator scheme will be given access to Azure AI Infrastructure options for high-end GPU virtual machine clusters. 

The move is aimed at supporting high-growth startups to accelerate the training and development of large language models (LLMs) and deep learning models (DLMs). 

Y Combinator startups will be the first to gain access following the update to Microsoft’s startup program. The firm said this will offer nascent businesses the “technical resources they need to quickly prototype and bring to market cutting-edge AI innovations”. 

“Y Combinator (YC) and its community of startup innovators will be the first to access this offering in private preview to a limited cohort,” said Charlotte Yarkoni, Microsoft’s president of commerce and ecosystems for cloud and AI.

“Microsoft Azure offers cloud-based scalable AI infrastructure, built for and with the world’s most sophisticated AI workloads, from delivering the largest and most complex AI models including GPT-4 and ChatGPT through Azure OpenAI Service to developers to infuse AI capabilities into many apps.”

In addition to this, startups will also be granted access to features and tools through Azure Machine Learning. The platform enables both low-code and code-based training of custom models, as well as the ability to fine tune frontier and open source models. 

Michael Seibel, managing director of Y Combinator, said the move comes in direct response to the increasingly large infrastructure requirements placed on firms training AI models. 

“With the overwhelming infrastructure requirements needed to do AI at scale, we believe that providing startups with high-performance capabilities tailored for demanding AI workloads will empower our startups to ship faster,” he said. 

Startups in Microsoft’s M12 venture fund will also be given access to dedicated supercomputing resources under the plans. 

Yarkoni said the long-term goal is to enable a broader variety of startups to access high-performance infrastructure and accelerate development.  

“Over time, our vision is to partner with additional startup investors and accelerators, with a goal of working with the ecosystem to lower the barrier to training and running AI models for any promising startup,” she said 

Startup ecosystem engagement

Tech giants such as Microsoft, Google, and AWS have developed ever closer relationships with high-growth startups specializing in AI in recent months. 

Google and AWS, for example, have pledged billions of dollars in support for Anthropic in a bid to draw upon the firm’s expertise and foundation models. 

Microsoft’s relationship with OpenAI has seen it invest around $10 billion dollars so far. This close relationship has resulted in a slew of product innovations and the integration of generative AI tools across all of Microsoft’s core services.  

With this latest move, Microsoft appears intent on drawing on the capabilities of high-growth AI startups and bringing them under the umbrella of its own cloud infrastructure. 

Yarkoni said collaboration with startups that are “leading innovation” in the AI space will be critical to ensuring long-term success. 

A recent survey from KPMG found that 75% of US-based CEOs believe generative AI is now a top investment priority. 

83% of respondents also said they anticipate an increase in generative AI investment by more than 50% over the next year. 

“For startups, this represents a once-in-a-generation opportunity to bring groundbreaking impact to a market hungry for change,” Yarkoni said.