Artificial intelligence has given rise to some of the most significant opportunities, leaving tech giants commanding premium valuations in the market. Microsoft and Google have reported a significant uptick in their cloud revenue due to integrating artificial intelligence into their offerings. Likewise, Meta has bounced back to growth in its lucrative advertising business, attributed to AI enhancing engagement levels and targeted advertising on its platforms.
Tech AI Expenditure
The growth run rate that the companies are enjoying has come at a price. The companies must spend billions of dollars to accelerate the development and launch of various AI-powered products and solutions. Microsoft has already pumped close to $13 billion into OpenAI as it sought to get its hands into AI solutions, including AI-powered chatbot ChatGPT.
Likewise, the software giant has confirmed that it spent $14 billion on capital expenditures in the recent quarter. It now expects the capital expenditures to increase significantly, driven by its ever-increasing AI infrastructure expenditures. Microsoft is not the only company experiencing a significant uptick in spending attributed to AI investments.
Alphabet says it registered a 79% increase in expenditure in the recent quarter to $12 billion and expects the same to increase for the rest of the year as it increases its focus on AI opportunities. Meta expects its capital expenditure for the entire year to increase by up to 42% and could range between $35 billion and $40 billion. The increase would be due to aggressive investment in AI research and product development.
Meta announcements caught investors by surprise, resulting in the stock edging lower. A point of concern is that social networking expenditure is increasing faster than sales growth.
Soaring AI Costs
The higher-than-expected expenditures among tech giants stem from AI models getting more extensive and more expensive to develop. As the demand for AI services increases, companies are forced to build more data centers to handle the vast troves of data needed to support various AI models.
While most of the costs stem from building foundations needed to support the various AI systems, the companies also incur additional costs once the systems are up and running. The best and most expensive AI models, including ChatGPT, are powered by large language models fed massive amounts of data. The push for more sophisticated artificial intelligence systems that can outperform humans on various tasks and the costs of developing such systems escalate even further.
The companies must procure more data to support the large AI models. In addition, they will have to spend more on acquiring computing power and training AI systems. According to Anthropic chief executive officer Dario Amodei, the current AI model crop costs $100 million to train. On the other hand, he believes the more powerful models in development that are expected to come out later in the year could cost up to $100 billion.
Much of the growing costs of developing and training AI models stems from the costs of chips needed to support the models. Not only are the Graphics Processing Units required to support AI models in short supply, but they are also exceedingly expensive. A standard chip for training AI models from Nvidia costs upwards of $30,000.
