Will the cloud eat your AI?

Comment: The major cloud providers have a lot to do when it comes to artificial intelligence. Does that mean these cloud providers are the only game in town?

Image: Krongkaew / Moment / Getty Images

More on artificial intelligence

“Leave all hope, you who come in here” was the caption Dante read as he passed through the gates of hell. Apparently, this is also true for anyone other than the major cloud providers when it comes to artificial intelligence, according to an analysis by Bain & Company. The CSPs [cloud service providers] is best placed because of the significant lead they have in using AI on a large scale, “said the report’s authors. Given that FirstMark investor Matt Turck recently called on how well startups have fared in the giants of the cloud, it is worth diving deeper into the forces the clouds bring to AI.First lots of data.

SEE: AI on the high seas: Digital transformation revolutionizes global shipping (free PDF) (TechRepublic)

AI-rich get richer

“CSP’s cloud and digital services have given them access to the vast amounts of data required to effectively train AI models,” the authors concluded. Such economies of scale have been an asset for cloud providers for years. Many years ago, RedMonk analyst Stephen O’Grady highlighted the “relentless economies of scale” that cloud providers brought to hardware – they could simply build cheaper than any business could hope to replicate in their own data centers. Now the CSPs enjoy a similar benefit with data.

But it’s not just a matter of raw data.

The CSPs also have more experience using this data on a large scale. The CSPs have products (e.g. Amazon Alexa to help with natural language processing or Google Search to help with recommendation systems). Lots of data feeding of ever smarter applications that feed more data into the applications … it’s a self-reinforcing cycle.

SEE: Digital transformation: A CXO’s guide (free PDF) (TechRepublic)

Oh, and the hardware mentioned earlier? The CSPs also have more experience in adjusting hardware to process large-scale machine learning loads. According to Bain, more than 15% of the major CSP’s servers based on conversations with CSP employees are focused on AI computing workloads, which may increase to more than 30% by 2025.

All this experience, in turn, leads the CSPs to dive deeper and deeper into increasingly sophisticated commercial AI models. As the report shows, the complexity of their deep-learning models has more than doubled every three to four months (Figure A).

Figure A

deep-learning-models.jpg

Photo: Bain & Company

And the CSPs do not gather this expertise. In fact, CSPs like Google (TensorFlow) and Facebook (PyTorch) have released these projects as open source, providing on-ramps to help drive the demand for their platforms. Brookings Institution Fellow Alex Engler pointed to this trend, saying that “for Google and Facebook, the open procurement of their deep learning tools (TensorFlow and PyTorch, respectively) may have [the effect of] further anchoring them in their already fortified positions. ”

And finally people. I’m not sure how Bain calculated these statistics, but the report suggested that the number of AI staff on five US CSPs (Amazon, Microsoft, Alphabet, Facebook, and IBM) exceeds the size of the AI ​​workforce in the next 45 U.S. companies combined. Again, it is unclear how to appoint an “AI employee”, but even though it is far away, it is not hard to believe that the CSPs tend to attract many talented engineers / others who are skilled in machine learning / other aspects of AI (and can afford to pay their not insignificant salaries).

SEE: Research: Video conferencing tools and cloud-based solutions dominate digital workspaces; VPN and VDI less popular among SMEs (TechRepublic Premium)

Back to Dante. Is all hope lost? Will the CSPs dominate ML / AI?

If so, no one seems to have informed startup entrepreneurs and the VCs that fund them. As Turck pointed out, “[T]the pace of innovation is too explosive in space for things to remain static for too long. Founders launch new startups, Big Tech companies create internal data / AI tools and then open them, and for every established technology or product, a new one seems to emerge every week. “A host of AI-related startups were announced over the past year, including C3.ai, UiPath, SentinelOne, Recursion and Darktrace. These companies cover security, pharmaceuticals and more.

So yes, the clouds have significant advantages in AI, but no, that does not mean they will dominate. More likely, many companies will benefit from the tools / services they provide, although the market remains open to other participants to help developers build machine learning and other elements of AI in their applications.

Disclosure: I work for MongoDB, but the views expressed herein are mine.

Also see

Leave a Comment