Microsoft’s AI Chief Says It Built a Better Model Than OpenAI’s With a Quarter of the Computing Power
Microsoft's AI chief told staff the company quietly trained a model that outperforms a key partner's flagship system at a fraction of the cost, but the market barely flinched, and the reasons why reveal a complicated bet hiding inside the…
Microsoft (NASDAQ:MSFT | MSFT Price Prediction) used its Microsoft AI unit on October 15, 2026, to announce a model that, if the internal numbers hold, could change how far its data center dollars go.
Microsoft AI said it had finished pre-training MAI-Base-2, a mixture-of-experts model (a design that switches on only part of the network for each task) with 80 billion active and 2.8 trillion total parameters. Microsoft AI CEO Mustafa Suleyman also reportedly told staff at an internal town hall that the model beats the one running OpenAI’s GPT-5.6 while using about a quarter of the computing resources OpenAI used.
The stock rose 1.08% to $539.44 on October 15, then fell 0.47% to $536.88 after hours. For a company worth $4.01 trillion, that was a quiet response, and the market appears to be waiting for proof.
What Microsoft Disclosed and What It Only Claimed
Microsoft said it trained the model on 16k GB200 graphics chips over 38 days in August and September 2026.
Microsoft AI said the model is 4-12x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base, the best open base model. FLOPs count raw calculations.
The rest of the announcement describes plans. The model “is currently undergoing post-training,” the stage that turns a raw model into a usable product, and Microsoft plans to “license our model weights to trusted partners through Foundry.” It gave no release date, pricing, or public benchmarks.
The OpenAI comparison comes only from Suleyman’s reported remarks, which rest on private tests of how well a model predicts patterns. Suleyman also claimed MAI-Base-2 used 12 times less computation than DeepSeek’s comparable system, and he said Microsoft was “set free” to pursue its own frontier ambitions.
Why Compute Efficiency Matters More Than the OpenAI Comparison
Microsoft spent $115.95 billion on capital projects in fiscal 2026. If it can reach frontier quality on a fraction of the compute, each dollar supports more training and more customers, raising the return on that spending.
Ownership also matters inside Copilot. Every query Microsoft answers with its own model is one where it keeps the margin it would otherwise pay a partner. On the July 29, 2026 call, CFO Amy Hood called model diversification “a margin improvement opportunity.”
The return is likely limited by scale, because Microsoft’s AI business already ran above $37 billion a year as of fiscal Q3 2026, so one model works on costs over time and is unlikely to swing earnings soon.
Where the OpenAI Claim Runs Into Trouble
Best pre-training can fade during post-training, and OpenAI could ship something best before MAI-Base-2 reaches customers, while OpenAI has not publicly confirmed or challenged the comparison and Suleyman runs the unit being judged.
Partner tension is likely. Microsoft holds a stake in OpenAI Group PBC reportedly valued at $135 billion, remains its primary cloud partner, sells its models on Azure, runs them in Copilot and has rights to its technology through 2032. Claiming publicly that it beat a partner that drives Azure demand creates friction.
Satya Nadella set this direction in July, saying “every model is substitutable.” MAI-Base-2 continues that strategy.
What Investors Should Watch Next for Microsoft Stock
This announcement likely strengthens Microsoft’s long-term position, although its core cloud business remains the main driver of the stock.
The company now has a large OpenAI stake that keeps paying off through Azure, plus a credible in-house option that limits OpenAI’s bargaining power, and licensing the model through Foundry would add to what Azure already sells, because Microsoft takes in revenue whichever model a customer picks.
That mix of a profitable core business and freedom to switch models likely sets Microsoft apart among large AI companies.
If the upcoming earnings call passes without management giving a Foundry release window or public MAI-Base-2 benchmarks, the OpenAI comparison will likely remain a talking point, and the stock will likely keep trading on Azure growth.
Contact [email protected] for any questions or corrections.
Omor Ibne Ehsan is a writer at 24/7 Wall St. He is a self-taught investor with a focus on growth, cyclical, and dividend equities that have strong fundamentals, value, and long-term potential. He also has an interest in high-risk, high-reward investments such as penny stocks.
