Wix CEO Avishai Abrahami has been vocal about Base44 being the future of the company’s AI strategy, and Q2 2026 results suggest that confidence is well-placed. With revenue hitting $563 million — up 15% year over year — and bookings reaching $569 million (a 12% annual increase), the numbers tell a clear story: Base44’s trajectory is no longer speculative. It’s operational.
As a researcher who spends most of her time studying LLM architecture decisions and their downstream economic effects, I find the Base44 story particularly compelling — not because it’s flashy, but because it’s structurally interesting. Let me explain why.
Why Base 1 Matters Beyond the Headline
The launch of Base 1, Base44’s proprietary large language model, is the kind of move that separates companies treating AI as a feature from those treating it as infrastructure. When you rely on third-party model APIs — OpenAI, Anthropic, Google — you’re essentially renting intelligence. Your margins are permanently coupled to someone else’s pricing decisions. Your latency is bounded by their infrastructure. Your differentiation is limited to prompt engineering and orchestration.
By building Base 1 in-house, Base44 made a deliberate architectural choice: own the inference layer. The immediate result, as Wix’s Q2 reporting confirms, is improved cost control and expected margin expansion. But the longer-term implications are more significant from a technical standpoint.
The Margin Math of Proprietary Models
Let’s think about what “improved cost control” actually means in this context. When you operate your own model, several cost variables shift in your favor:
- Inference cost per query drops — you can optimize the model for your specific use cases, pruning capacity you don’t need and accelerating paths you use constantly.
- Batching and caching become first-class operations — rather than paying per-token to an external API, you control how requests are grouped and served.
- Fine-tuning cycles tighten — feedback loops between user behavior and model improvement become internal, reducing the lag between data collection and model updates.
For a platform like Base44, which presumably handles high volumes of structured generation tasks (website creation, design decisions, content production), even modest per-query savings compound rapidly at scale. That’s the margin upside Wix is signaling to investors — not a one-time gain, but a structural shift in unit economics.
A Technical Researcher’s Perspective on the Risks
I want to be measured here. Building a proprietary LLM is expensive, and maintaining one is arguably more so. The AI model space moves fast. Foundation model capabilities improve quarterly. A proprietary model that’s competitive today could lag behind frontier systems within months if the team doesn’t sustain investment in training data, architecture updates, and evaluation infrastructure.
The question I’d want answered — and that Wix hasn’t publicly detailed — is how Base 1 is positioned relative to general-purpose frontier models. If it’s a smaller, task-specific model optimized for Base44’s domain (web creation, layout generation, content structuring), that’s a defensible strategy. Domain-specific models can outperform larger general models on narrow tasks while costing a fraction to run. If it’s attempting to be a general-purpose competitor, the capital requirements become much steeper.
My bet, based on the cost-control framing, is the former. And if so, that’s a smart architectural decision.
What This Signals for the AI Platform Economy
Wix’s Q2 results, powered by Base44’s performance, represent a broader pattern I’m tracking across the industry. Companies that initially integrated AI through API partnerships are now selectively building proprietary models for their highest-volume, most margin-sensitive workflows. They’re keeping API access for edge cases and long-tail tasks where building custom solutions doesn’t justify the cost.
This hybrid approach — own your core inference, rent your periphery — is becoming the template for AI-native platform economics. Base44 appears to be executing on exactly this playbook.
The Numbers in Context
Fifteen percent revenue growth and twelve percent bookings growth are solid figures for a company of Wix’s scale. The fact that these results are attributed specifically to “strong Base44 performance and continued core Wix growth” suggests the AI division isn’t cannibalizing the legacy business — it’s additive. That’s the dual-engine story investors want to hear.
From where I sit, the technical decision to build Base 1 is the more consequential data point than any single quarter’s revenue number. It tells us Base44 is building for durable margin advantage, not just top-line growth. And in the current AI economy, that distinction matters enormously.
🕒 Published: