Keep pulling the thread on Arvind Krishna.
In 2017, IBM concluded that competing in the public cloud market would require an annual investment of $5 to $10 billion just to remain the fifth-largest player, making it an unattractive investment.
IBM acquired Red Hat to gain a portfolio that would enable it to partner with all major cloud providers rather than compete with them directly.
IBM spun off its IT services business, which represented one-third of its workforce, because it was declining at 5% annually and did not align with the company's new focus on innovation, high margins, and growth.
Arvind Krishna believes the current AI infrastructure build-out is ahead of what the market can support in the next few years.
The committed build-out of over 100 gigawatts for AI data centers implies a $6 to $8 trillion capital investment in semiconductors.
To achieve a five to seven-year payback on the current AI infrastructure build-out, an additional $1 to $2 trillion of annual revenue is needed, which Arvind Krishna believes is not currently available.
Arvind Krishna predicts that the largest AI models will become a commodity with low switching costs, which will prevent them from sustaining high margins with a significant moat.
Arvind Krishna predicts that only two or three companies, rather than a half-dozen, will be able to sustainably build and operate the largest AI models.
Arvind Krishna predicts that AI agents will replace the front-end of much software, which will decrease the total value of those products.
The IBM Z17 mainframe has post-quantum cryptography built directly into the platform.
Arvind Krishna believes quantum computing could become a business worth hundreds of billions of dollars for IBM.
Quantum computers pose a national security threat due to their ability to break most of today's encryption standards using Shor's algorithm.