Ever since the advent of generative artificial intelligence (AI), many experts have highlighted the potential problems with this technology, ranging from misinformation to the destruction of humans through superintelligence AI. However, many failed to see the most critical problem with AI, one that is proving to be destructive to its own growth – the economics of scale. As per reports, major tech firms are struggling to turn their AI businesses into profit despite the massive excitement from consumers as it is very costly to keep AI products operational.
According to a report by The Wall Street Journal, AI tools which are unproven in terms of market durability, can be quite expensive to run as they require robust servers with expensive chips that consume lots of power. In particular, the cost related to power is very tricky since AI does not work like other software products. Answering each query will require the AI to run a specific algorithm which can use up more energy depending on how complicated the query is. That means heavy users in particular can push the cost to the company quite high.
We are now on WhatsApp. Click to join.
As per the report, Microsoft is losing a significant amount of money on its GitHub Copilot, an AI assistant that helps coders create, fix, and translate codes. More than 1.5 million users have used it so far, and have paid $10 a month for the service. However, the company is actually losing a whopping $20 per user per month, and some users were even costing the company as much as $80, the report quoted unnamed sources as saying.
Tech companies trying to find solution to the AI problem
These are untested waters, so pricing troubles were expected. However, it appears companies were not prepared for the costs to run so high, because just like other software, they expected profits to be generated as more people subscribed to the service, but that is not possible with AI, at least as long as computation remains energy inefficient.
Yet, companies are finding their own unique ways to combat the situation. For example, Google and Microsoft have both raised the AI feature-infused services for Microsoft 365 Office and Google Workspace. Both will be charging $30 a month for the service whereas the non-AI counterpart costs $13 and $6 a month, respectively.
While Microsoft and Google have resorted to pushing the price, Zoom, the video conferencing platform, has taken another route. The company has multiple AI models to offer its features to users. For most of the tasks, such as summarizing meetings and composing chat messages, it uses inexpensive AI models meant for specific tasks and thus saves on costs.
Adobe is using a system of credits to prevent the company from being burdened by losses. Every user who opts for the AI features gets a fixed number of credits that can be used to run a query and to get it to help the user with the project. Once all is spent, the speed slows down significantly to discourage users. The same method is also used by Dall-E, Midjourney, and Stable Diffusion.
“We are trying to provide great value but also protect ourselves on the cost side,”Adobe CEO Shantanu Narayen told the WSJ.
But if this trend continues, and the AI excitement is not met with the kind of returns investors are hoping for, a lot of the VC money might disappear next year, which will affect a lot of startups and new-age companies.
One more thing! HT Tech is now on WhatsApp Channels! Follow us by clicking the link so you never miss any updates from the world of technology. Click here to join now!