OpenAI Declares “AGI Era” With Astra Launch. Anthropic’s Fable 5.1 More Than Doubled Its Predecessor’s Science Score. MSFT and AMZN Both Win, but One Has A Slight Edge

OpenAI President Greg Brockman ended a September 3 briefing on GPT-6 Astra with “Welcome to the AGI era,” after saying he personally believed OpenAI had reached AGI. The benchmark results explain the swagger. OpenAI reports that Astra scored 64.6% on Terminal-Bench Science 0.1, ahead of Anthropic’s new Claude Fable 5.1 at 52.6% and GPT-5.6 Sol at 22.4%. Anthropic’s own evaluation puts Fable 5.1 at the same 52.6%, more than double its 24.7% reproduction of Fable 5. The companies’ numbers come with different harness details and standard errors, but the central point survives: Fable made a gigantic leap on the benchmark, but Astra still finished ahead.

For investors, however, Microsoft and Amazon now have substantial exposure to both labs. What we’re interested in is which company has the stronger monetization stack around those relationships.

Microsoft Corporation (NASDAQ:MSFT) began rolling Astra through Microsoft Foundry on September 3, where Fable 5.1 is also available. Its amended OpenAI agreement keeps Azure as OpenAI’s primary cloud partner, gives Microsoft a nonexclusive license to OpenAI models and products through 2032, continues OpenAI revenue-sharing payments through 2030, and preserves its equity exposure. Anthropic has meanwhile committed to buy $30 billion of Azure capacity, while Microsoft committed to invest up to $5 billion in Anthropic.

That gives Microsoft several monetization routes: Azure consumption, Foundry distribution, OpenAI licensing and revenue sharing, equity exposure, and first-party products such as Copilot. Azure and other cloud services revenue grew 43% in the June quarter, while Microsoft 365 Copilot passed 30 million paid seats. Microsoft can therefore push frontier AI through Microsoft 365, GitHub, Dynamics, and security products rather than relying primarily on infrastructure demand.

Hedge funds were less enthusiastic before the launches, however. Insider Monkey’s database shows 273 funds holding MSFT at the end of Q2 2026, down from 282 in Q1, although Arrowstreet Capital raised its share position by 14%. That decline is a genuine caution flag, but of course it precedes Astra.

The tradeoff is in the cost. Microsoft Cloud gross margin fell to 65% as Azure mix, AI infra, and usage costs weighed on profitability. MSFT expects roughly $175 billion of calendar-2026 capital spending after a lease-accounting reclassification, while OpenAI also has greater freedom to use rival clouds.

Amazon.com, Inc. (NASDAQ:AMZN) has built an equally serious two-lab strategy. Anthropic still names AWS as its primary cloud and training provider and has committed to spend more than $100 billion on AWS technologies over ten years while securing up to five gigawatts of capacity, including Trainium chips. AMZN invested another $5 billion in Anthropic and could invest up to $20 billion more if milestones are met.

Amazon also agreed to invest $50 billion in OpenAI. OpenAI and AWS expanded an existing $38B infrastructure agreement by another $100 billion over eight years, while OpenAI expects to consume roughly two gigawatts of Trainium capacity. AWS also became the exclusive third-party cloud distribution provider for OpenAI Frontier, and the companies are developing customized OpenAI models for Amazon products.

AWS therefore has massive infrastructure exposure to both labs. Sales rose 37% to $42.2 billion in the June quarter, producing $16.6 billion of operating income. Hedge-fund sentiment strengthened as well: 369 funds owned AMZN at the end of Q2, up from 353, while Arrowstreet lifted its position by 24%.

The burden is found in the capital intensity. Amazon’s trailing free cash flow swung to a $7.6 billion outflow as property and equipment purchases increased by $66.1 billion, primarily because of AI investment. Neither lab relationship is widely exclusive, either.

All in all, that leaves two legitimate winners. Amazon may have the stronger pure infrastructure case through AWS, Trainium, and contracted demand from both labs. Microsoft gets the narrower edge because it can monetize frontier AI further above the infrastructure layer through OpenAI licensing and revenue-sharing rights and direct distribution across its enterprise software ecosystem.

In conclusion, both can profit from the model race. Microsoft just has more ways to monetize what happens after the models leave the data center.

READ NEXT: NVIDIA (NVDA): What Foxconn and Super Micro Are Telling Us about the AI Boom  and Pony AI Is Scaling Robotaxis Fast—Can the Stock Reach BofA’s $17 Target?

Follow Insider Monkey on Google News.