Microsoft AI’s MAI-Transcribe-2 undercuts OpenAI, Google and ElevenLabs on price and speed
THE SO WHAT
At $0.10 per hour, high-quality speech-to-text is effectively a rounding error—any audio-heavy workflow that isn’t transcribed and indexed is now a choice, not a cost constraint. Teams building call analytics, meeting tools, or media platforms should assume transcription is free and compete on downstream insight, not raw ASR.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIGoogle is quietly building an AI version of Canva inside Workspace — and Nano Banana could make it seriously useful
AI-native design tools inside Workspace collapse a whole class of “lightweight creative” SaaS into something users get by default. If you sell marketing, slideware, or design tools, assume Google will own casual creation — your survival depends on deep workflows, brand governance, and integrations, not templates.
Applied AITCL’s New Phones Can Convert E-books Into Audiobooks and Podcasts Using AI
On-device AI that turns any e-book into an audiobook or podcast blurs the line between reading and listening — and bypasses traditional audio licensing. Publishers and platforms need to revisit rights models and UX assumptions fast, because users will expect every text asset to be instantly “listenable.”
Applied AIThe AI visibility gap: Why great brands disappear from AI answers
If AI assistants are answering questions directly, traditional SEO metrics are lagging indicators at best. Brand and growth teams need an explicit "AI surface" strategy this quarter—structured content, data partnerships, and evals on how often you show up in assistant answers for your category.
Applied AINvidia Buying AI Platform Hugging Face in $13 Billion Deal
A $13 billion acquisition plus a $1 billion retention pool for Hugging Face is Nvidia turning the open-source AI commons into a first-party distribution channel for its silicon. If your AI roadmap leans on HF models or datasets, assume tighter coupling with Nvidia’s stack and start mapping where you want that dependency versus where you need alternatives.