Subvolts take IBM bets on small, efficient AI models, prioritizing data quality and Generative Computing over massive scale. Apr 19, 2026 - Developers
- Learn about IBM's technical stack for efficient AI and the concept of Generative Computing for programmable AI elements.
- Business
- Understand IBM's strategy for enterprise AI using smaller, safer, and more sustainable models, and how data quality is key.
This episode features a key IBM executive discussing the company's core AI strategy, offering a distinct perspective on model development.
#335 Sriram Raghavan: Why IBM Is Betting Everything on Small AI Models
Audio streams from Craig S. Smith. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
In this Eye on AI episode, Craig Smith interviews IBM's Sriram Raghavan about the company's focus on small AI models. Raghavan explains IBM's decision to train its Granite models directly using reinforcement learning, prioritizing data lineage, safety, and efficiency for hybrid cloud environments. The discussion covers how data quality surpasses model size as a competitive advantage and introduces IBM's concept of Generative Computing.
IBM is building smaller AI models instead of huge ones. They believe focusing on the quality of data used to train these models, and new ways to program them, will lead to better and safer AI for businesses.
This episode offers insight into IBM's distinct strategy in the AI race, diverging from the industry trend of massive models. It's valuable for enterprise leaders and AI developers exploring efficient, safe, and sustainable AI solutions. Listeners will understand how IBM aims to compete with frontier labs by focusing on data quality and novel architectures like Generative Computing, potentially influencing the direction of enterprise AI development and deployment.
- HostCraig S. Smith
- GuestSriram RaghavanVice President of AI, IBM Research
- Guest: Sriram Raghavan, Vice President of AI at IBM Research
- Title: #335 Sriram Raghavan: Why IBM Is Betting Everything on Small AI Models
- Show: Eye on AI
- Length: 60 minutes
- IBM's approach: Trains Granite models directly using reinforcement learning, not distilling from larger models.
- Performance claim: 8 billion parameter model matches GPT-4o and Claude 3.5 on code and math benchmarks via inference time scaling.
Why IBM Is Betting Everything on Small AI Models
In this episode of Eye on AI, Craig Smith sits down with Sriram Raghavan, Vice President of AI at IBM Research, to explore one of the most important debates in enterprise AI right now. Do you actually need a massive model to get world class results? IBM's answer is no, and Sriram breaks down exactly why.
Sriram explains why IBM chose to train its Granite models directly using reinforcement learning rather than distilling from larger models like most of the industry. The reason goes beyond performance. It comes down to data lineage, safety alignment, and a belief that small, efficient models are the only sustainable path for enterprises running AI across hybrid cloud environments.
We get into the full technical stack behind that bet. How data quality has replaced model size as the real competitive advantage. Why parameter count is becoming the wrong metric entirely. How IBM's inference time scaling techniques allow an 8 billion parameter model to match the performance of GPT-4o and Claude 3.5 on code and math benchmarks. And why IBM is pioneering a new concept called Generative Computing, which treats AI models not as prompt receivers but as programmable computing elements with runtimes, modular LoRA adapters, and proper programming abstractions.
Sriram also shares where IBM Research is headed next, including breakthroughs in continuous learning, agent orchestration, and making unstructured enterprise data actually usable at scale.
Subscribe for more conversations with the people building the future of AI and emerging technology.
Stay Updated:
Craig Smith on X: https://x.com/craigss
Eye on A.I. on X: https://x.com/EyeOn_AI
(00:00) Why IBM Skips Distillation and Trains Small Models Directly
(04:50) Did We Even Need Giant AI Models in the First Place?
(08:12) How Data Quality Became the New Competitive Moat
(11:54) Why Parameter Count Is the Wrong Way to Measure a Model
(15:36) Reinforcement Learning Without Losing Broad Capabilities
(22:05) Inference Time Scaling: Getting Big Model Results From Small Models
(28:12) Generative Computing: Treating AI as a Programming Element
(36:40) Why IBM Open Sources and How Small Models Make It Sustainable
(41:25) The Path to Continuous Learning Without Rewriting Weights
(51:00) IBM's Full Roadmap: Models, Data, and Agents
- reinforcement learning
- Sriram Raghavan
- A way to train AI by giving it rewards for correct actions.
- distilling
- Sriram Raghavan
- Making a smaller AI model learn from a larger, more capable one.
- data lineage
- Sriram Raghavan
- Tracking where data comes from and how it's processed.
- LoRA adapters
- Sriram Raghavan
- Small, add-on pieces that can customize AI models without retraining the whole thing.
55 min42 min51 min34 minUkraine's Secret Weapon: The Points System Winning the Drone War | Andrii Hrytseniuk
WithAndrii Hrytseniuk
2 h 38AI in the AM: 99% off search, GPT-5.5 is "clean", model welfare analysis, & efficient analog compute
WithAnna Patterson, Lukas Petersson, Zvi Mowshowitz, Naveen Verma
57 min28 min
Questions people ask
- Why is IBM focusing on small AI models?
- IBM believes small, efficient models offer a sustainable path for enterprises, prioritizing data lineage, safety alignment, and performance in hybrid cloud environments.
- What is Generative Computing?
- Generative Computing treats AI models as programmable elements with runtimes and modular adapters, moving beyond simple prompt responses.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Craig S. Smith.