The prevailing trend of massive infrastructure investment in artificial intelligence is facing scrutiny from within the tech industry. IBM’s CEO, Arvind Krishna, has publicly questioned the scale of spending by companies racing to develop and deploy advanced AI services like OpenAI’s ChatGPT and Google Gemini.
Krishna, in an interview with the Handelsblatt, indicated that IBM has adopted a more conservative approach, deliberately choosing to avoid the significant capital expenditure seen by competitors. Companies like Alphabet, Google’s parent company, are reportedly planning investments of $75 billion this year alone.
Instead of building its own extensive infrastructure, IBM intends to prioritize a flexible leasing model. This strategy allows the company to adapt more readily to rapidly changing demands and technological advancements. Krishna’s projections suggest a dramatic shift in costs; he anticipates the expenses associated with inference – the practical application of AI models – could decrease to just one percent of current levels within five years.
IBM’s stance is particularly notable given its past role as a pioneer in AI with the Watson supercomputer. While the company initially led the field, it missed the current wave of large language model hype. Furthermore, cloud-based data center business is dominated by rivals such as Microsoft and Amazon Web Services (AWS).
To differentiate itself, Krishna is focusing on highly specialized AI models, emphasizing investment in numerous smaller, more targeted applications. These models are designed to operate on less powerful hardware. IBM is also offering access to competitor models through strategic partnerships. Krishna highlighted the significant untapped potential within the corporate sector, estimating that 99 percent of enterprise data remains unanalyzed by existing large language models.
