Microsoft Dumps OpenAI in Its Apps to Save Cash with MAI
Microsoft spent billions backing OpenAI, only to realize paying someone else for compute is a great way to go broke. Now the tech giant is quietly kicking Sam Altman's models out of its ecosystem to cut costs.
The software behemoth announced two new homegrown artificial intelligence models through Azure AI Foundry: MAI-Image-2.5-Pro for visual generation and MAI-Voice-2-Flash for speech synthesis. Instead of renting expensive external algorithms forever, the tech giant built its own tools to handle complex image edits and ultra-fast voice responses in high-load systems.
The financial math behind the shift is brutal. Generating images with MAI-Image-2.5-Pro costs $5 per million text input tokens and $106 per million output image tokens. Meanwhile, the new voice synthesizer works twice as fast as its predecessor and dropped costs by 32 percent down to $15 per million characters.
The true motive becomes crystal clear inside the company's own product suite. Bing Image Creator has completely transitioned to the new image model, while PowerPoint reduced its rendering expenses by a whopping 84 percent compared to using GPT-Image-2.
User engagement metrics also got a handy corporate boost. In OneDrive, saved generated images grew by 26 percent alongside a 25 percent reduction in latency, proving that cheaper models do not necessarily anger the customer base.
The voice side tells an identical story of relentless corporate trimming. Dynamics 365 Contact Center cut operational computing expenses by 89 percent using the Flash voice model, while Dragon Copilot switched to MAI-Transcribe-1.5, cutting speech recognition errors in half across 58 languages.
Nothing solidifies a tech partnership quite like replacing the partner's entire product line behind closed doors. Watching corporate giants build billions in alliances only to cannibalize them for margin optimization remains the ultimate silicon valley romantic story.
Source: Azure AI Foundry
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.