Bottleneck Capital
Long/Short Investor, Technology, Media & Telecom
Microsoft's Copilot profit line is metered agent work rather than seat sales, and that meter could outrun the next 10 million seats. Over 30 million paid Copilot seats are distribution: at the $30 monthly list price they annualize to $10.8bn before discounts, and Microsoft doesn't break out Copilot revenue. Usage-billed Autopilot credits then meter the work each seat already unlocked. x.comÂ
Microsoft's 30M+ paid M365 Copilot seats could become a market for metered agent work. Its new Cowork, Code and Autopilot tools use usage-based billing on long-running tasks, while tenant identity, sandboxed hosting, audit controls and admin spend limits may get enterprise buyers past security reviews. It can also standardize collaboration across Fortune 500 departments instead of individual Claude or Codex setups. These are promises; watch repeat usage, cost per task and budgets. x.com x.comÂ
Microsoft's new Copilot could make it over-earn if executed well, much like Meta's Muse, and all its compute will come in handy. Per @dee_bosa, Copilot is Microsoft's bet that the AI race is moving from models to products, choosing among models while keeping customers inside Microsoft, with Autopilot working across users' apps and data. A revenue share from OpenAI adds further upside. x.comÂ
Microsoft's Copilot demand is outrunning supply, with companies reportedly hitting Copilot credit limits because they want to consume more. @SouthernValue95 notes Microsoft disclosed $37B of AI ARR in April 2026, growing triple digits and larger than Anthropic, OpenAI or AWS AI run-rates, and customer checks show it leads AI for large enterprises across Azure, Office, cyber and data software. x.comÂ
Microsoft is losing ground in fast AI innovation and is not cheap enough to compensate. Copilot was strong a few months ago but is no longer competitive in an agent-driven world, and the stock closed below its 200-week moving average for the first time since 2013. At about 21x forward EPS for 22% growth it is solid but not cheap, and compared with Nvidia there is little argument for allocating even 1% of a portfolio to Microsoft. x.comÂ
Microsoft looks set to break out again, this time on agents after its last breakout came from the cloud rerating. Nadella is making Copilot the operating system for work, unifying chat, coding and long-running agents in one entry point instead of scattering them across Studio, Cowork and GitHub Copilot, to defend against OpenAI and Anthropic selling all-in-one apps to its core enterprise customers. Pricing shifts from seats toward compute, with Cowork, Code and Autopilot billed by usage. x.comÂ
Microsoft's good-enough approach to AI offers a model beyond capex that eats free cash flow before returns are visible, and its earnings may have triggered an inflection in the AI trade. Related stocks rose as soon as the CFO said Microsoft will be free-cash-flow positive next year. Copilot is not the best AI product, but it has 30 million paid seats and keeps improving. Only time will tell which approach is right, but good enough usually works out for Microsoft. x.com x.comÂ
Microsoft is a great business but still an expensive bet at $385, because heavy AI capex is eroding free cash flow. Capex was revised to $190B for the year, $43B above expectations, and free cash flow fell to $15.8B. Revenue grew 18% to $82.9B, operating income 20%, Azure 40% and AI run rate reached $37B, with a 47% operating margin. The average analyst target of $560 implies 45% upside only if the AI spending converts on schedule. x.comÂ
Microsoft's Copilot could win in enterprise AI long term because it runs natively in Microsoft 365, where enterprise work already happens, with only about 30 million of some 450 million paid M365 seats on it so far. Microsoft guided above consensus, and Azure growth improved from 39% to 43%, with 45% guided. Value is shifting from models to infrastructure like public clouds. OpenAI, 27% owned by Microsoft, is the biggest concern, which is partly why Microsoft is exploring open-source models. … Show more
Microsoft's 50% Azure growth and prospect of positive cash flow in 2027 make the stock look bulletproof. Headcount cuts, lower Anthropic spend, a better Copilot routed to cheaper tokens, and coming proprietary chips and model usage add to the case. It should get Apple's multiple plus recognition for cybersecurity. x.com x.comÂ
Microsoft's $75B Activision Blizzard deal has been a total failure, underperforming cash. Three years in, Game Pass has about 30M subscribers at $20 a month, roughly $7.2B a year; at a 33% margin that is $2.4B, a 30-year payback, and the deal didn't materially help console sales. The same $75B in a high-yield savings account would throw off over $3B a year. x.com x.comÂ
Microsoft's valuation can recover if a few conditions come through, and the current risk-reward looks acceptable. A jury dismissed Elon Musk's claims against OpenAI and Microsoft on procedural grounds, removing a feared overhang despite a planned appeal. Microsoft 365 Copilot paid seats grew 33% in a quarter to 20M, about 4.3% penetration, and the new $99 E7 bundle builds Copilot into the base package. Maia cannot end its Nvidia reliance soon, but a rotation to cash-rich value stocks could help. … Show more
