🕶️ NVIDIA's startup sells cheap AI

Good morning, good morning, good morning!

This is the Friday edition where we learned that "too cheap to matter" just crossed a billion in revenue. Fireworks built a $17.5B company on the models no one brags about, Uber is financing its own regulatory workaround, and somewhere a procurement team finally got leverage over an AI vendor.

The economics of building are getting quietly weird in useful ways.

Most teams scale customer service by hiring, but with ElevenAgents, real-time voice and chat agents handle high volumes of interactions and resolve them faster.

When routine requests come in, they get answered in the moment, in 70+ languages. However, the complex ones route to humans with full context, so your team focuses only on the conversations that need judgment. The experience holds up because the agents sound human, not robotic. And with Expressive Mode, they read tone and adapt - de-escalating, reassuring, and guiding each conversation to a clear resolution.

You can train every agent on your knowledge base, SOPs, and policies, deploy across voice, chat, and WhatsApp, then keep improving with real-time CSAT tracking, A/B testing, and Guardrails.

The outcome is the one leaders care about: higher resolution rates, higher CSAT, and support that scales with demand instead of headcount. Pricing is transparent and flat at $0.08 per minute, with no add-ons to work around.

Get started free

Fireworks just raised $1.5 billion at a $17.5 billion valuation. The company hit $1 billion in annualized revenue — a 5x year-over-year surge — by making AI inference so cheap that closed-model pricing starts to look like a subsidy for inefficiency.

How cheap: Fireworks processes 40 trillion tokens per day across GPUs from 20+ suppliers, serving clients like GitLab and MongoDB at 5-10x less than frontier alternatives. A new Microsoft Foundry partnership and plans to triple headcount to 600 suggest this isn't a scrappy cost play anymore. It's a full infrastructure bet against the vertically integrated clouds.

The deeper signal: Amazon and Google bundle inference with their own proprietary models, creating lock-in at premium price points. Fireworks' multi-supplier approach flips that — teams can swap models without switching clouds. As inference costs crater, the moat for frontier labs narrows to raw model quality alone, and that's a moat that erodes every time an open-weight model closes the gap.

For founders building on AI: the cost of coupling your stack to a single model provider is rising fast. Design for fungibility now, or pay the switching tax later.

Uber just agreed to buy Delivery Hero for $14.8 billion. The last independent global food-delivery platform is gone, absorbed into a single network spanning 99 markets and $236 billion in combined gross bookings.

The price itself tells the story. Two months ago, Uber's €33 offer sat below the previous close. Large holders pushed for €40. The final number — €41.50 per share — landed €1.50 above even that holdout line, a 26% premium that bought board endorsement and an irrevocable commitment from Prosus, locking Uber's economic interest at roughly 53%.

But the structure matters more than the headline. To clear regulatory overlap in 14 markets, Delivery Hero will sell those units to SSW Partners for $1.6 billion — financed mostly by Uber itself. The acquirer is bankrolling its own divestiture. If regulators approve that mechanism, every future roll-up architect gets a new playbook for buying scale while shedding overlap.

For founders in consolidating sectors, the signal is straightforward: modular, cleanly separable business units just became a deal-closing asset.

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