IBM CEO Arvind Krishna walked onto the Masters of Scale stage this week and told founders something most AI vendors will not say aloud: Your Year 1 AI ROI will almost certainly be negative. That is not pessimism. That is data. IBM is tracking $5 billion in cumulative efficiency gains against a 2022 baseline — but getting there meant accepting an upfront loss, redesigning workflows from the ground up, and treating Year 1 as an infrastructure investment, not a returns play. Krishna''s framework is simple. Year 2 is where the 10x returns show up — but only for organisations that stop bolting AI onto broken processes and start reimagining the work itself. He calls it day zero thinking, and most enterprises never get there because their CFO kills the project before the compounding begins. The episode also makes a case against the foundation model arms race. IBM is betting on AI orchestration — stitching models together into agentic workflows — not building bigger language models. That is a contrarian bet worth watching. Here is the hard question for every business leader watching: Are you measuring your AI programme on a one-year horizon when the real returns do not materialise until year two? — Priya Sharma Full video link https://youtu.be/9YoyMz28ojA