Microsoft is boosting capacity to support OpenAI’s GPT-4-5, GPT-5 models

Engineers at Microsoft are boosting infrastructure capacity in preparation for the latest iterations of OpenAI’s large language models, the first of which could be implemented by the end of February.

Sources close to Microsoft informed Tom Warren of The Verge that the software giant is planning to host OpenAI’s newest GPT-4.5 as early as next week.

OpenAI CEO Sam Altman recently disclosed that the company plans to “next ship GPT-4.5,” but didn’t provide an exact date. 

Although GPT-4.5 has been described as a “mid-generation” update, OpenAI has reportedly been training the model with synthetic data, which, according to IBM, can overcome data scarcity when training and fine-tuning AI models. OpenAI’s use of synthetic data to train GPT-4.5 was initially reported by The Information.

In addition to expecting GPT-4.5 any day now, The Verge’s source said Microsoft expects to receive the more powerful GPT-5 in late May. 

Altman described GPT-5 “as a system that integrates a lot of our technology, including o3,” which refers to OpenAI’s latest reasoning model. On Jan. 31, the company released a smaller o3 model called o3-mini.

Microsoft currently hosts OpenAI’s models on its Azure platform. However, Microsoft clarified that the service does not interact with any tools operated by OpenAI, including ChatGPT.

Microsoft and OpenAI expanded their partnership last month through President Donald Trump’s $500 billion AI venture called Stargate. 

OpenAI has also “made a new, large Azure commitment that will continue to support all OpenAI products as well as training,” Microsoft said.  

Related: OpenAI’s newest ChatGPT agent can do ‘deep research’ online

AI race heats up

Since launching in November 2022, ChatGPT has set records as the world’s fastest-growing consumer software platform. By February 2025, it had amassed roughly 400 million weekly active users, which marked a 33% jump in less than three months, according to OpenAI’s chief operating officer Brad Lightcap.

This remarkable growth has allowed OpenAI to seek funding at a $340 billion valuation, according to CNBC.

However, competition is heating up with the recent launch of DeepSeek, an open-source AI model with Chinese origins that was developed at a tiny fraction of ChatGPT’s cost. 

The launch of DeepSeek threatened the prevailing paradigm that OpenAI — and the United States — would remain the dominant player in the AI market. 

Venture capitalist Marc Andreessen dubbed DeepSeek as “AI’s Sputnik moment,” which refers to a society’s realization that it needs to play catch up with rapid technological developments made elsewhere in the world.

Markets certainly reacted with that collective awe as tech stocks, Bitcoin (BTC) and the broader cryptocurrency market plunged in the wake of DeepSeek’s release.

Magazine: Train AI agents to make better predictions… for token rewards