The probability of achieving Artificial General Intelligence (AGI) with current technologies like LLMs is extremely low (0-1%); true AGI will require new approaches like neuro-symbolic AI.
The United States must lead in quantum computing for national and economic security, and currently holds a fragile two-year lead over China.
The most profitable strategy in AI is not to build the largest, commoditizable models, but to provide smaller, efficient models and tools for specific, high-value enterprise use cases.
The current multi-trillion dollar AI infrastructure build-out is economically unsustainable in the short term, as the required incremental revenue to justify the investment does not yet exist.
IBM's future lies in a focused portfolio of hybrid cloud, AI software, and consulting, having deliberately exited the direct public cloud competition and divested its lower-growth legacy services business.
2017
Krishna states IBM leadership concluded that competing in the public cloud market was an unattractive investment, requiring $5-10 billion annually just to remain the fifth-largest player, setting the stage for a strategic pivot.
2018
IBM announced its acquisition of Red Hat, a move Krishna describes as foundational to its new strategy of partnering with, rather than competing against, major cloud providers.
Post-2020
As CEO, Krishna oversees the spin-off of IBM's IT services business (Kyndryl), a unit he notes was declining at 5% annually, to sharpen the company's focus on innovation, high margins, and growth in hybrid cloud and AI.
2025-2026
Krishna details rapid progress in quantum simulations, from a 5-atom molecule in June 2025 to a 300-atom molecule in November 2025, and a 12,000-atom molecule (half of the protein trypsin) by April 2026.
Mid-2026
Krishna publicly reaffirms that IBM is on track to achieve its ambitious goal of demonstrating quantum advantage by the end of 2026 and expects 'incredible returns' from quantum and cybersecurity investments within 2-3 years.
▶Pragmatic AI Strategy: Enterprise Focus over Frontier Models
Arvind Krishna outlines a distinct AI strategy for IBM that deliberately avoids the race to build the largest 'frontier' models. He believes these models will become commoditized with low switching costs and that current LLM technology has a near-zero chance of achieving AGI. Instead, IBM focuses on smaller, cost-effective models like its Granite family (under 100B parameters) and applying AI to specific enterprise use cases, such as M&A synergy and code generation.
Krishna is positioning IBM to be a picks-and-shovels provider and a specialized solutions expert in the AI gold rush, betting that sustainable, high-margin business lies in enterprise application rather than in the capital-intensive and potentially low-margin competition of building foundational models.
▶Quantum Supremacy as a National and Corporate ImperativeJul 2026
Krishna consistently frames quantum computing not just as a technological frontier but as a critical issue of economic and national security, highlighting a two-year U.S. lead over China. He has set an aggressive public timeline for IBM to achieve 'quantum advantage' by the end of 2026 and deliver a large-scale fault-tolerant machine by 2029, viewing the technology as a potential 'hundreds of billions' dollar business for IBM.
By tying IBM's quantum progress directly to U.S. competitiveness, Krishna elevates the company's role from a commercial vendor to a strategic national asset, likely aimed at securing government support and positioning IBM as the indispensable leader in a field with profound geopolitical implications.
▶IBM's Strategic Reinvention for a Hybrid Cloud World
Krishna details a deliberate corporate transformation, marked by the spin-off of its declining IT services business and the pivotal acquisition of Red Hat. This strategy was born from the 2017 conclusion that competing in public cloud was a losing battle. The new IBM is structured around high-margin software (50%), consulting (33%), and specialized hardware (20%), with a core thesis that enterprises will overwhelmingly adopt a hybrid, multi-cloud strategy.
Krishna's strategy is a candid admission of IBM's past strategic errors and a decisive pivot to a platform-agnostic enabler role, aiming to capture value across the entire cloud ecosystem rather than competing head-on with hyperscalers.
▶Sober Economics of the AI Infrastructure Boom
While bullish on AI's long-term impact, Krishna provides a cautious analysis of the current market dynamics. He quantifies the announced AI data center build-out at 100 gigawatts, representing a potential $8 trillion in capital expenditure. He questions the market's ability to generate the $1-2 trillion in new annual revenue needed to justify this investment on a 5-7 year payback, suggesting the build-out is ahead of demand.
Krishna's analysis suggests an impending market correction or consolidation where capital has been misallocated, indicating that IBM's more measured approach to capital expenditure and focus on immediate ROI may prove advantageous.