According to a report covered by the Register, Microsoft is losing up to $80 per user per month on its GitHub Copilot services. While the service costs $10 a month, heavier users cost the company as much as $80 every 30 days. Microsoft’s strategy of running products at a loss aims to build a dedicated user base and potentially increase prices in the future. The cost of running AI models like Copilot is high due to expensive hardware and electricity consumption. Microsoft is investing heavily in AI to secure a first-mover advantage in the market. It is speculated that Microsoft may reveal its custom AI accelerator to control the cost of AI operations.
This was reported in the Wall Street Journal too. Tech companies like Microsoft, Google, and Adobe are struggling to turn the hype around AI into profits. Generative AI tools are expensive to operate and require powerful servers and chips. Companies are experimenting with pricing and charging models, such as higher price tags, monthly caps, and consumption-based charges. The high costs and complexities of AI products make it challenging to generate profits. Still, companies are hopeful that profitability will improve as computing costs come down and AI technology advances.
According to McKinsey, generative AI can reduce cloud migration efforts by 30% to 50% when implemented correctly. Using large language models (LLMs) can help organizations understand system infrastructure, identify weaknesses and strengths, and assess the success of the migration. Generative AI has symbiotic potential with the cloud, accelerating migration and unlocking its benefits. The four primary use cases for generative AI are content generation, customer engagement, synthetic data creation, and coding. LLMs are expected to remain in hyper-scale environments for the next five to six years.
Google is encouraging its partners, particularly system integrators and independent software vendors, to embrace generative AI (GenAI) through its partner program updates. The company has announced increased support, funding, and incentives for partners who align with its focus areas, including gen AI. Several companies, such as Accenture, Automation Anywhere, and Deloitte, have already signed up for the GenAI partner initiative. Google also introduced new expertise designations and plans to double rewards for partners who drive workload adoption and accelerate the adoption of strategic products.
Why do we care?
As AI providers better understand economies of scale, prices for AI features could increase. We’re all intrigued because this won’t be free – AI is expensive – and many vendors aren’t sure how they’ll make money yet. However, with tangible benefits like migration cost savings, it’s not a matter of if the math will be figured out but when,
The McKinsey report cites the importance of guardrails for regulated industries and protecting proprietary data. Tailored or massive models are not always necessary, and an API gateway can help monitor access to non-proprietary models or data. And all of that work to get customer guidance will pay off.