Highlights:

  • Despite its widespread adoption, GitHub Copilot has become a significant financial drain, as indicated by an individual familiar with the financial data.
  • Adobe Inc. has adopted a distinct strategy, implementing a credit-based system for its AI image generation tool, Firefly.

The surge in interest in new generative artificial intelligence technologies was initiated by the introduction of ChatGPT towards the end of last year. Nowadays, nearly every major software provider has embraced this trend.

Despite the widespread excitement for generative AI among both technology providers and customers, there remains a challenge in monetizing these potent new products. Contrarily, according to a recent report by the Wall Street Journal, companies like Microsoft Corp., Google LLC, and OpenAI LP are believed to be incurring substantial financial losses due to providing generative AI capabilities to their users.

An early example of a generative AI product introduced is GitHub Copilot. Programmers utilize this service for tasks such as generating, correcting, and translating software code. Microsoft, the parent company of GitHub, has stated that GitHub Copilot boasts over 1.5 million users and approximately half of the generated code is written by these users.

Despite its widespread adoption, GitHub Copilot has become a significant financial drain, as indicated by an individual familiar with the financial data. GitHub charges users USD 10 per month for access to Copilot, but on average, they incur a loss of about USD 20 per month per customer.

Generative AI proves costly due to the extensive time needed for model training, fine-tuning, and the ongoing need for substantial resources to operate them daily, even after the initial training and refinement phases.

In numerous scenarios, there is also an aspect of excessive capability that exceeds what is necessary. For example, ChatGPT operates using OpenAI LP’s GPT-4 model, regarded as one of the world’s most potent models. However, many ChatGPT enterprise subscribers utilize it for highly specific and limited tasks. As stated by the Journal, employing GPT-4 to summarize an email is akin to delivering a pizza using a Lamborghini.

Cheaper Models at Higher Prices

Many companies are considering developing less powerful models suited for simpler business tasks to address the financial strain. Additionally, some are contemplating raising their prices. For instance, Microsoft plans to introduce an AI-infused version of its Office 365 software suite, charging an additional fee of approximately USD 30 per month. The least expensive version of Office 365 is approximately priced at USD 10 per month. The AI capabilities will include composing emails and PowerPoint presentations, generating Excel spreadsheets automatically, and undertaking various other tasks. Likewise, Google is considering requesting users to pay USD 30 a month for access to the generative AI features within its productivity software. In contrast, its lowest subscription tier presently stands at just USD 6.

Simultaneously, reports from other sources suggest that Microsoft is actively working on the development of less powerful models to address the issue of excessive capabilities. For instance, Microsoft is creating smaller, cost-effective AI models designed explicitly for Bing, focusing solely on web search functionality. According to reports, certain models will be based on open-source AI technology from companies like Meta Platforms Inc.

Adobe Inc. has adopted a distinct strategy, implementing a credit-based system for its AI image generation tool, Firefly. The Firefly service will slow down if a user exhausts their allotted monthly credits, according to the company, a way to deter abuse.

Customers and Investors still Curious

Microsoft, Google, and other companies contemplating higher charges for their AI services face a delicate balancing act. Not everyone is convinced that the software justifies the additional cost.

Currently, the challenges in monetizing generative AI haven’t deterred investors. This year, they have collectively invested billions of dollars in the most promising startups. According to The Journal, OpenAI is discussing a potential share sale that could elevate its valuation to over USD 90 billion. This represents a threefold increase from its value at the beginning of the year.

Nevertheless, a significant portion of the industry believes it’s just a matter of time before investor enthusiasm diminishes. When that moment arrives, many individuals will scrutinize the expenses associated with AI operations and seek profitable approaches for utilizing this technology.