Orbit 64
Every December, Rob Toews publishes ten AI predictions in Forbes, then comes back a year later and grades himself in public. That habit of putting a stake in the ground, and owning the misses, makes him one of the more interesting voices in the space. In this conversation with Jon Wulkan, he explains why the frontier has narrowed to a handful of Western labs, why even well-resourced challengers have struggled to keep pace, and why the much-feared "SaaS apocalypse" is probably overdone. They also dig into the economics underneath the hype: why model prices are likely to rise once the big IPOs land, why one of the three remaining frontier players has a structural funding edge the others can't match, and why an obscure accounting question about chip depreciation could reprice the entire AI infrastructure trade. Along the way, Toews reaches for a comparison that sticks: we're building data centers that draw twice the power of San Francisco, while the human brain runs on 20 watts. His advice for vertical software CEOs? Hire for slope, not y-intercept..
Inside the episode
- There are only three true frontier AI labs left in the West - OpenAI, Anthropic and Google DeepMind. Meta and xAI have both fallen behind despite enormous capital, and it's becoming a durable top tier that's extremely hard to break into.
- AI model prices are set to rise, not fall. Once OpenAI and Anthropic IPO and face pressure to become profitable, expect pricing to normalise upward - with Google's self-funding core business giving it a structural advantage the other two don't have.
- An obscure accounting question could reprice AI infrastructure. With new chip generations now arriving every 12–18 months, five-to-six-year depreciation schedules look increasingly hard to justify - and shorter schedules would make today's data centre economics look far less attractive.
- The "SaaS apocalypse" is overdone. Frontier labs will chase the biggest horizontal markets, but hard-won customer context is a real, durable moat for the vast majority of vertical software companies they'll never fully reach.
Transcript

Jonathan Wulkan
Welcome to Orbit, the Hg podcast series, where we talk to tech leaders and investors and hear how they've built and backed some of the most successful technology companies in the world. I'm Jon Wulkan and today I have the pleasure of speaking to Rob Toews. Rob is a partner at Radical Ventures and an AI columnist at Forbes.
His annual prediction piece comes with a public scorecard: he publishes both his calls and his marks, five out of ten in 2025 and six in 2024. Rob brings a wealth of experience and perspective, including as an advisor to the Obama White House on autonomous vehicles, and as a leading thinker on physical AI before it was even a field.

Rob Toews
Thanks for having me.









