$124B Data Problem: How Synthetic Data Accelerates AI

Sep 18, 2025

Balint Pasztor, CEO of Diffuse Drive and former robotics engineer at Bosch, tackles the intricate challenges of data scarcity in AI. He explains how synthetic data can revolutionize the training of autonomous systems, condensing years of data collection into mere hours. Balint dives into the significance of edge cases for safety, the workings of diffusion models for realistic data generation, and a unique three-step approach to optimize AI performance. His insights are essential for anyone interested in the future of robotics and autonomous technology.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Three Pillars Determine Autonomy

Autonomous systems need three pillars: hardware, AI models, and high-quality data.
Missing any pillar cripples performance and deployment.

ANECDOTE

Bosch Startup Sparked The Problem

Balint describes an internal Bosch startup where data scarcity blocked progress despite ready models.
That experience motivated Diffuse Drive to solve data shortages for autonomy.

INSIGHT

Edge-Case Data Is Safety-Critical

Lack of rare edge-case data prevents safe deployment and can cause harm.
Machines lack human intuition and must see examples of rare events to generalize safely.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

When your AI has to make decisions in the real world, the data you don’t have can hurt people.

DiffuseDrive CEO Balint Pasztor joins IT Visionaries to unpack the $124B data scarcity problem holding back autonomous systems — and how synthetic data (done right) can compress years of data collection into hours.

We dig into: why edge cases matter (and how to safely create them), closing the sim-to-real gap with diffusion models, preventing model drift, and building a data engine for physical AI across defense, robotics, and industrial automation. If you care about self-driving cars, drones, QA on the factory floor—or just shipping AI that survives the messiness of reality—this one’s for you.

Key Moments:

00:00 Introduction to Autonomous Driving Challenges

00:26 Meet Balint Pasztor and Diffuse Drive

01:14 The Importance of Synthetic Data

06:39 The Role of Synthetic Data in AI Training

18:07 Understanding Diffusion Models

23:28 Challenges in Real-World Data Collection

26:46 Three Steps to Improve AI Performance

32:13 Overcoming Non-Obvious Data Challenges

36:00 Balancing Data Quantity and Quality

40:28 Future of Autonomous Systems and Physical AI

This episode of IT Visionaries is brought to you by Meter - the company building better networks. Businesses today are frustrated with outdated providers, rigid pricing, and fragmented tools. Meter changes that with a single integrated solution that covers everything wired, wireless, and even cellular networking. They design the hardware, write the firmware, build the software, and manage it all so your team doesn't have to.

That means you get fast, secure, and scalable connectivity without the complexity of juggling multiple providers. Thanks to meter for sponsoring. Go to meter.com/itv to book a demo.

---

IT Visionaries is made by the team at Mission.org. Learn more about our media studio and network of podcasts at mission.org.

Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.