The terminal's grip is slipping. Old infrastructure is either getting expensive new challengers or getting skipped entirely , Bloomberg under API pressure, LLMs under existential doubt, and a slow trickle of enterprise bets that say the tools founders take for granted may not survive their own moats.

This week asks whether ambition means scaling what works or betting that it won't. The founders picking sides are raising real money either way.

Attio is the agentic CRM that runs the work behind every win. It meets you where you work, compounds every customer signal into context, then acts on it so you can move fast, stay sharp, and scale without breaking.

Then Ask Attio to:

30,000+ teams use Attio to drive more revenue at scale.

Try Attio Now

The market data terminal is finally getting an API-native antagonist. Databento just closed an oversubscribed $97M raise led by NEA, taking total funding to ~$127M and explicitly positioning its platform as a Bloomberg replacement rather than another niche feed vendor.

The bet: programmatic access becomes default, terminals become UI skins. Databento writes its own handlers for 70+ exchanges, serves 3,000+ firms, grew revenue 6.65x YoY, retained 97% of enterprise customers, and hit profitability with fewer than 30 engineers running 20+ PB of data, a capital-efficiency profile more like a great devtools company than a Wall Street vendor.

The new money goes to expanding asset coverage, building out Europe and APAC, and scaling to 20+ data centers and 100+ PB of storage. If Databento wins, market data becomes another API primitive founders can treat like Stripe, not a terminal contract.

HDR was supposed to make videos look better.

Instead, it gave us blinding Instagram photos, washed-out videos, broken color grading, and endless arguments about why the same image looks completely different across devices.

So what happened?

In this episode, Caya dives into the hidden history of HDR, color management, Apple’s ColorSync system, sRGB, Dolby Vision, and the decades-old decisions that shaped the modern internet.

In a market drunk on token counts, Yann LeCun just ordered water. He’s leaving Meta to build AMI, a $1B bet on “world models”, systems that learn physics and causality from video, not just text; arguing that scaling LLMs will never cross the gap between chatbots and house cats.

The wager is that the real frontier is self-supervised learning over unlabelled sensory data, closer to how a child builds intuition about gravity, motion, and objects. AMI’s JEPA-style models compress scenes into abstract representations and reason there, instead of hallucinating the “next pixel.”

Near-term, that points toward industrial control, robotics, and complex systems rather than consumer assistants. Longer-term, LeCun’s federated Tapestry proposal; shared models trained on sovereign data silos, offers an alternative to both closed US stacks and politically fragile Chinese ones.

Scale your growth: Join names like HubSpot, Ahrefs, and Deel. Put your product in front of 85,000+ startup founders. Let's talk.