Tuesday Sep 22
GOVERNANCEMicrosoft

Microsoft Exec Called AI Scraping Theft

22SEP
93% FEWER CLICKSTHE MEMOCOPILOT

Unsealed filings in the Times lawsuit show what Microsoft and OpenAI said in private. One Microsoft director called training on news the largest theft of labor in history. Nadella said paywalled content should be licensed.

Brent Hecht, Microsoft's director of Applied Science, wrote the theft line in a January 2023 memo. A 2024 Microsoft deck showed Copilot cutting Times click-through by up to 93 percent and called it a doom loop.

Nick Turley, who runs ChatGPT, said publishers face an existential threat and that the products are largely substitutive. Greg Brockman replied ah nice to a colleague's Times paywall hack.

The data: 91,692 Times and Daily News works in training, over two million Times documents via Common Crawl, copyright notices stripped. OpenAI has not commented.

full brief & sources

⚡ Why this matters

  • Fair use cases turn on market harm. Microsoft's own deck quantified the harm at 93 percent.
  • The Trump administration filed a brief supporting fair use on September 2. These memos are the plaintiffs' answer.
  • Every AI company has internal emails like these. Discovery is now the real risk in every copyright suit.

🔍 What happened

  • TechCrunch's Rebecca Bellan and the Washington Post reported the unredacted filings on September 17. The quotes come from the New York Times' brief; the underlying exhibits remain sealed.
  • Brent Hecht, Microsoft director of Applied Science, in a January 2023 memo: training on news content is "an astonishing theft of unprecedented proportions" and "the largest theft of labor in human history."
  • A January 2024 Microsoft deck found Copilot's answer engine cut Times click-through by up to 93 percent versus Bing and called it a "doom loop" where the product threatens its own suppliers.
  • Satya Nadella testified that paywalled content "should be licensed," which would have required OpenAI to retrain.
  • Nick Turley, head of ChatGPT, said publishers face an "existential threat" and the products are "largely substitutive." Greg Brockman said models are "excellent at news."
  • The Times says 91,692 of its works and the Daily News' appeared in mid-training data, plus over 2 million nytimes.com documents via Common Crawl. Project Taxi and Mango added 160,903 works with copyright notices stripped.

💬 Smart takes

  • Steven Lieberman, Daily News counsel: "they knew that what they were doing was wrong."
  • Brent Hecht's memo, per the filing: it is "highly unusual that an end-product threatens the economic foundations of its essential suppliers."
  • Skeptic: we are reading the plaintiffs' selection of quotes from sealed exhibits. Context could soften the memos, and internal dissent is not a legal admission.

🧭 Where this goes

  1. LikelyMicrosoft and OpenAI settle with the Times and Daily News before a jury sees these slides.
  2. Likelyother publishers in pending suits file to unseal similar internal documents within months.
  3. Possiblethe court admits the 93 percent click-through deck as evidence of market harm, weakening the fair use defense across cases.
  4. Wild CardCongress uses the memos to fast-track a compulsory licensing scheme for news content.

🥄 The Spoon Take

Forget the legal theory. Read the deck. Microsoft measured that its own product cut Times traffic by 93 percent and called it a doom loop, then shipped it. Fair use is decided on harm to the market. The defendants just documented the harm themselves. Every AI lab's discovery folder looks like this.

🤔 Pushback

These are cherry-picked lines from sealed exhibits, and a court can find internal worry without finding infringement.

·TechCrunch - Unredacted filings revealed·MLex - NYT v OpenAI filings·Washington Post - Court records show
Friday Aug 14
FLOPSMicrosoft

Microsoft Cuts Copilot's Failed Features

14AUG
RIGHT TO EXISTMICROSOFTRETIRED

Two years after calling AI a generational shift, Microsoft is cleaning house. The consumer and business Copilot apps are merging, and Group Chats, AI podcasts, Deep Research, and the Mico mascot are gone.

The internal verdict was blunt. Jacob Andreou, the EVP in charge, wrote in a July memo that the app must earn 'the right to exist.' Whatever failed that test got cut.

Users lose four features by August 18, including the floating-blob character widely read as AI-era Clippy. Paying enterprise customers keep a tool called Researcher. Files migrate to OneDrive.

Everyone is consolidating. Claude folded Cowork into Chat, OpenAI folded Operator into ChatGPT. The AI super-app era is really a subtraction era.

full brief & sources

⚡ Why this matters

  • Microsoft was the loudest 'AI changes everything' voice. Its consumer AI app just admitted feature bloat.
  • The whole industry is consolidating AI apps into single surfaces. This is the clearest example yet.
  • Feature kills show what users actually ignored, which is better market data than launch announcements.

🔍 What happened

  • Aug 13 - Microsoft confirmed it is merging the consumer Copilot app and the business Microsoft 365 Copilot app.
  • Group Chats, AI-generated podcasts, Copilot Labs, and consumer Deep Research shut down by August 18.
  • Mico, the animated Copilot character launched in 2025, is retired.
  • Files from the standalone Copilot app migrate to OneDrive.
  • The Information reported a July memo from EVP Jacob Andreou saying Copilot must earn 'the right to exist.'

💬 Smart takes

  • Jacob Andreou, Microsoft EVP: Copilot needs to earn 'the right to exist' in customers' lives.
  • Microsoft, officially: the goal is a 'simpler, more cohesive experience.'
  • Sarah Perez, TechCrunch: it's 'an admission that Copilot has lost its way.'
  • Skeptic: pruning features is what healthy products do. The flop framing may say more about AI expectations than about Microsoft.

🧭 Where this goes

  1. Likelymore Copilot feature kills follow before the merged super-app ships.
  2. LikelyGoogle and Meta run the same consolidation play on their AI app sprawl within two quarters.
  3. Possiblethe merged Copilot gains real consumer share once it stops confusing users with two apps.
  4. Wild CardMicrosoft revives a Mico-style character once AI companions become a proven category.

🥄 The Spoon Take

The real story isn't the dead features, it's the scoreboard. Microsoft shipped everything and let users vote. Podcasts, group chats, and mascots lost. A focused chat-plus-work surface won. Expect every AI roadmap to shrink the same way.

🤔 Pushback

Calling this a flop is easy. Killing weak features two years into a new category is normal product hygiene, and Copilot still ships on a billion Windows machines.

·TechCrunch - Copilot feature cuts·GeekWire - Copilot app merger·Microsoft - Copilot blog
STRATEGYMicrosoft

Microsoft Quietly Shrinks In China

14AUG
15 DOORS SHUTMICROSOFT1.5%

Big Tech's China exit is happening in slow motion. Reuters reports Microsoft closed or left at least 15 branches and ventures; China is down to 1.5 percent of revenue. One door stays open.

Microsoft considered a fuller exit in 2023, then chose controlled exposure. Top researchers were relocated and new research hubs built outside China.

What stays is tied to AI and cloud. Azure still serves Chinese companies expanding overseas, a market US restrictions leave open.

The model is neither in nor out. Expect other US platforms to copy the shape: minimal footprint, selective revenue.

full brief & sources

⚡ Why this matters

  • The biggest US software company is showing what decoupling actually looks like in practice.
  • AI export controls are redrawing where research and revenue can live.
  • China strategy is now a template question for every global platform company.

🔍 What happened

  • Reuters detailed the retreat on Aug 13.
  • Microsoft has closed or withdrawn from at least 15 branches and joint ventures in recent years.
  • China's share of Microsoft's worldwide revenue fell to roughly 1.5 percent by 2024.
  • The company weighed a more complete exit in 2023 before keeping selected operations.
  • Top researchers were relocated and alternative research hubs built outside China.
  • Azure and AI tools still serve Chinese businesses expanding abroad.

💬 Smart takes

  • Reuters: what remains is increasingly tied to AI, cloud, and Chinese companies going global.
  • Skeptic: 1.5 percent of revenue means China was never the prize for Microsoft - the real decoupling test belongs to Apple and Tesla.

🧭 Where this goes

  1. Likelymore US tech firms formalize the minimal-footprint China model within a year.
  2. LikelyChinese firms going overseas remain the one growth segment US clouds can serve.
  3. PossibleBeijing restricts foreign cloud services for outbound Chinese companies in response.
  4. Wild Carda US-China AI deal reopens research collaboration before 2028.

🥄 The Spoon Take

Decoupling talk usually sounds like an on-off switch. Microsoft shows it's a dimmer, turned slowly over years. The endgame isn't leaving China - it's shrinking to a shape no regulator on either side finds worth attacking.

🤔 Pushback

A slow retreat by a company with tiny China revenue proves little about whether hardware-dependent giants could ever do the same.

·Reuters - Microsoft China retreat·Tech Startups - Aug 13 briefing
Sunday Aug 9
STRATEGYMicrosoft

Microsoft's AI Revenue Is Mostly OpenAI

9AUG
$24.1 BILLIONMICROSOFTOPENAI

The AI boom has fewer customers than it looks. Microsoft booked $24.1 billion from OpenAI last fiscal year. That is roughly 70% of everything it calls AI revenue.

Bloomberg pulled the figure from regulatory filings. One customer supplies more than 7% of the company's entire top line. Satya Nadella had guided investors toward $37 billion in annual AI sales back in March.

Nothing here is hidden or improper. Every dollar is disclosed. But Microsoft funded OpenAI, OpenAI spent it on Azure, and Microsoft logged the result as growth.

If you size this market from cloud earnings, you are partly counting two labs spending investor money. Worth knowing before you build a forecast on it.

full brief & sources

⚡ Why this matters

  • The clearest public number yet on how concentrated AI revenue actually is.
  • Enterprise buyers sizing the market off cloud earnings are reading a partly circular figure.
  • Concentration risk now sits inside the largest software company's growth story.

🔍 What happened

  • Bloomberg reported that Microsoft disclosed $24.1 billion in OpenAI revenue for the fiscal year ending June 2026.
  • That is roughly 70% of Microsoft's AI revenue, per Bloomberg's analysis.
  • It is also more than 7% of Microsoft's total fiscal 2026 revenue.
  • The figure may include Azure compute, revenue share on OpenAI products, and other commercial agreements.
  • Satya Nadella said in March the company was pacing toward $37 billion in annual AI revenue.

💬 Smart takes

  • Ed Zitron, writer at Where's Your Ed At: "Microsoft disclosures suggest that OpenAI is about 70% of AI sales. Wow!"
  • Windows Central: that level of concentration on a single customer reads as unhealthy for a business this size.
  • Skeptic of the skeptics: every platform business starts concentrated. AWS was mostly Netflix and Amazon before it was everyone else.

🧭 Where this goes

  1. Likelyanalysts start asking every hyperscaler to break out AI revenue by customer.
  2. LikelyMicrosoft leans harder on Copilot seat numbers to show demand outside OpenAI.
  3. PossibleOpenAI shifts more compute off Azure and Microsoft's AI growth rate visibly slows.
  4. Wild Carda regulator treats the investment-to-revenue loop as a disclosure question, not just an accounting one.

🥄 The Spoon Take

The AI demand story and the AI supply story are the same story right now. Two labs are buying most of the compute, funded largely by the people selling it. That does not make it fake. It does make it fragile.

🤔 Pushback

Concentration is normal early in a platform market, and Copilot seats plus Azure AI customers are growing outside the OpenAI line.

·Bloomberg - Microsoft's AI sales disclosure·Where's Your Ed At - The 70% breakdown·Neowin - $24.1B from OpenAI
Wednesday Aug 5
AGENTSMicrosoft

Microsoft Open-Sources Agent Training

5AUG
67.5% SWE-BENCHAGENT GYMOPEN

Agents now get a gym before they get a job. Microsoft Research shipped Orchard, an open framework that trains agents in realistic environments before deployment. Training, not prompting, becomes the differentiator.

Orchard separates agent training from execution. Its Kubernetes-based environment collects training data, runs reinforcement learning rollouts, and evaluates agents at scale. Three recipes ship with it: software engineering, browser use, and personal assistants.

The numbers are real. Starting from a 30-billion-parameter Qwen model, Orchard-SWE hits 67.5 percent on SWE-bench Verified, a new open-source record for its size. Code and datasets are on GitHub and Hugging Face.

Analyst John Sviokla calls it the missing layer of the agent stack. If agents can be trained like employees, model choice matters less.

full brief & sources

⚡ Why this matters

  • Environment-aware training is emerging as the layer between models and deployed agents.
  • Open-source teams can now train deployment-grade agents without frontier-lab budgets.
  • Shifts enterprise agent quality from prompt engineering to training pipelines.

🔍 What happened

  • Microsoft Research released Orchard, an open framework separating agent training from execution.
  • Core is Orchard Env, a lightweight Kubernetes environment for rollouts, data collection, and evaluation.
  • Three recipes ship: Orchard-SWE, Orchard-GUI, and Orchard-Claw for coding, browser, and assistant tasks.
  • Orchard-SWE reaches 67.5 percent on SWE-bench Verified after supervised fine-tuning plus reinforcement learning.
  • Orchard-GUI posts 74.1 percent on WebVoyager, the strongest open-source result.
  • Full framework, recipes, and trajectory datasets released on GitHub and Hugging Face.

💬 Smart takes

  • John Sviokla, GAI Insights: Microsoft open-sourced 'the missing layer of the agent stack.'
  • Microsoft Research: the same infrastructure trains agents 'directly inside real deployment harnesses.'
  • Skeptic: enterprises don't want to train agents - they want to buy ones that already work.

🧭 Where this goes

  1. Likelyagent-training pipelines become standard enterprise practice alongside fine-tuning by mid-2027.
  2. LikelyOpenAI and Google ship competing open agent-training stacks within six months.
  3. PossibleOrchard-trained open models undercut proprietary coding agents on price for routine work.
  4. Wild Cardregulators start asking where your agent was trained - the gym becomes a compliance artifact.

🥄 The Spoon Take

The agent stack keeps growing new layers, and Microsoft keeps open-sourcing the plumbing. Models got commoditized, then frameworks. Now training environments. Whoever owns the gym where agents learn owns the quality bar, and Microsoft just handed everyone the same gym.

🤔 Pushback

Training agents demands data and MLOps maturity most enterprises lack - Orchard may stay a research toy outside the top teams.

·Microsoft Research - Orchard blog·arXiv - Orchard paper·GitHub - microsoft/Orchard·AI Business - Training cost angle
Saturday Aug 1
STRATEGYMicrosoft

Microsoft Posts Biggest Rally Since 2008

1AUG
$480B IN A DAYLAST YEARAZURE

Wall Street just picked its AI winner for the week. Microsoft stock jumped 15% and added roughly $450 billion in value. Azure cloud growth beat guidance, breaking Nvidia's own one-day record.

The single-day gain topped $450 billion, the largest ever recorded by any US company. The old mark belonged to a chipmaker, not a software firm.

CFO Amy Hood guided next quarter's growth to 45%, above the 41% analysts expected. Revenue from the cloud unit hit nearly $30 billion this quarter, up from $21 billion a year ago.

Investors read it as proof AI capex is finally showing up on the income statement. Every other hyperscaler's next earnings call just got a higher bar to clear.

full brief & sources

⚡ Why this matters

  • Largest single-day market value gain by any US company on record.
  • First hard proof this earnings season that AI infrastructure spend is converting into cloud revenue.
  • Raises the bar every other hyperscaler must clear next quarter.

🔍 What happened

  • Microsoft shares closed up more than 15% on July 30, 2026.
  • The move added roughly $450 to $480 billion in market value in a single day.
  • It broke Nvidia's previous single-day record of $441 billion, set in April 2025.
  • Azure revenue came in near $30 billion for the quarter, up from about $21 billion a year earlier.
  • CFO Amy Hood guided 45% Azure growth for the next quarter, above the 41% Wall Street expected.
  • Microsoft's total market cap closed near $3.35 trillion.

💬 Smart takes

  • William Blair analyst Jason Ader: Azure's growth sailed past the company's own guidance of 39% to 40%.
  • Skeptic: a single earnings pop doesn't prove AI capex pays for itself long term, especially with memory and chip costs still climbing.

🧭 Where this goes

  1. Likelyother hyperscalers face harder questions on their next call if their cloud growth misses Microsoft's bar.
  2. LikelyAzure's AI-driven growth narrative becomes the default template analysts measure every cloud vendor against.
  3. PossibleMicrosoft's rally cools once markets price in the higher expectations it just set.
  4. Wild Carda weak print from a rival hyperscaler next quarter triggers a broader AI-stock selloff.

🥄 The Spoon Take

One earnings call just answered the market's biggest AI question. Does the spending show up in revenue? For Microsoft, yes. That's the number every CFO defending an AI budget will point to next.

🤔 Pushback

One good quarter of cloud growth doesn't settle whether the industry's total AI capex will ever earn its cost of capital back.

·Fortune - Microsoft's $480B rally·Fortune - AI moat debate·Quartz - Microsoft record cap gain
Wednesday Jul 8
STRATEGYMicrosoft

Microsoft Launches Frontier Company

8JUL
MOVE INMICROSOFTCUSTOMER

Buying an AI model isn't the hard part anymore. Microsoft built a $2.5B unit with 6,000 engineers to make AI actually work. Three rivals made the same bet in eight weeks.

Commercial Business CEO Judson Althoff says this isn't just forward-deployed engineering. He calls it the industry's biggest outcome-driven engineering org.

The pitch: customers keep any model they want, Microsoft just makes it work. Four major labs made a similar move within eight weeks. AWS committed billions to the same idea days earlier.

The model race is quietly turning into a services race. Whoever staffs the rollout may end up owning the customer relationship.

full brief & sources

⚡ Why this matters

  • Enterprises have models but can't turn them into working systems alone.
  • Deployment services lock in the customer relationship longer than any API contract.
  • Four major labs made the same bet inside eight weeks - that's not a coincidence.

🔍 What happened

  • Jul 8: Microsoft launched Microsoft Frontier Company, a new operating unit.
  • $2.5 billion committed, 6,000 engineering and industry experts assigned.
  • Judson Althoff, Microsoft's Commercial Business CEO, says it goes beyond 'forward-deployed engineering.'
  • The unit redesigns workflows, deploys agents, and sets up governance after go-live.
  • Customers can keep using any model in the ecosystem, not just Microsoft's.
  • AWS and Microsoft together put $3.5 billion into similar deployment engineering this week.

💬 Smart takes

  • Judson Althoff: the unit will be 'the largest, most capable, outcome-driven engineering organization in the industry.'
  • Althoff: customers are 'in very different places right now, and trying to really figure out AI.'
  • Skeptic: a vendor-run deployment arm still nudges customers toward that vendor's own model and cloud.

🧭 Where this goes

  1. LikelyGoogle and Anthropic announce their own deployment-services arms within two quarters.
  2. Likely'AI deployment services' becomes a standard line item on enterprise software budgets.
  3. Possibleindependent AI consultancies lose ground as labs offer deployment for free with a subscription.
  4. Wild Cardone of these deployment units gets spun out as its own company within 3 years.

🥄 The Spoon Take

The model race just became a side quest. Whoever's engineers sit inside your company owns the relationship, not whoever built the model. Four labs reached that conclusion within eight weeks. That's a market signal, not a coincidence.

🤔 Pushback

Forward-deployed engineering is expensive consulting with a new name, and services margins never match software margins.

·TechCrunch - Frontier Company launch·Microsoft - Frontier Company details·GeekWire - $2.5B AI deployment unit
Tuesday Jun 30
STRATEGYMicrosoft

Copilot Bills Jump To $3,000

30JUN
$3,000/MOCOPILOTTHE BILL

GitHub Copilot moved to usage-based billing on June 1. June 30 is the first full cycle close. Developers are posting bills that jumped from $50 to $3,000 in heavy agentic work.

The flat $29 plan is gone for heavy users. Agentic workflows burn tokens fast. One developer's bill went from $29 to $750. Another hit $3,000.

This is the real cost of letting agents run. Every autonomous step is metered. The cheap-coding-agent story just met its invoice.

Budgets now need an agent-runtime line. Teams will cap concurrency or route simple tasks to cheaper models. Watch usage-based pricing spread across every AI tool.

full brief & sources

⚡ Why this matters

  • First full billing cycle reveals the true cost of agentic coding at scale.
  • Usage-based pricing turns AI coding from a fixed cost into a variable one.
  • Forces teams to budget agent runtime like cloud compute.

🔍 What happened

  • Jun 1: GitHub Copilot switched to usage-based billing.
  • Jun 30: first full monthly cycle closes.
  • Developers report bills jumping from $29 to $750.
  • Heavy agentic workflows hit $3,000 a month.
  • Each autonomous agent step consumes metered tokens.

💬 Smart takes

  • Developers on social: projected costs jumped 10-100x versus the old flat plan.
  • Skeptic: the screenshots are worst-case heavy users, not the median developer who still pays little.

🧭 Where this goes

  1. Likelyteams add agent-runtime caps and budgets this quarter.
  2. Likelymore AI dev tools move to usage-based pricing.
  3. Possiblea market opens for cost-routing that sends simple tasks to cheap models.
  4. Wild Carda backlash pushes a vendor back to flat-rate pricing as a differentiator.

🥄 The Spoon Take

The agentic coding demo was free. The production bill is not. Usage-based pricing exposes what autonomous agents actually cost when they run all day. The next FinOps fight is over agent-hours, not seats. Teams that don't cap concurrency will get surprised.

🤔 Pushback

Heavy-user screenshots make scary headlines, but most developers run light and still pay less than a flat seat would cost.

·The Register - GitHub metered billing·BuildFastWithAI - Jun 30 cost reports·BuildFastWithAI - Jun 29 AI news
Tuesday Jun 23
PRODUCTMicrosoft

Microsoft's AI Coworker Bills Per Task

23JUN
PAY PER TASKFORTUNE 5001 CENT EACH

Microsoft's AI coworker, Copilot Cowork, is now live worldwide. It runs long multi-step tasks and bills by usage, not a flat seat. Half the Fortune 500 already use it. It runs on Claude.

Copilot Cowork hit general availability June 16. It finishes complex work end-to-end, not just drafts.

Billing is usage-based, in Copilot Credits, at one cent each on pay-as-you-go. Price per task depends on model, context, tools, and runtime. It is off by default, with spend limits.

It runs on Anthropic's Opus 4.8 and Sonnet 4.6 today. Microsoft's own test claims it is 30-40% cheaper than Claude's Cowork. A cheaper in-house model, Cowork 1, is coming.

full brief & sources

⚡ Why this matters

  • Metered agent billing arrives at enterprise scale, not just in dev tools.
  • Microsoft sells an AI teammate to over half the Fortune 500 on day one.
  • Microsoft's flagship agent runs on Anthropic's models, then undercuts Anthropic on price.

🔍 What happened

  • Copilot Cowork went generally available worldwide on June 16.
  • It requires a Microsoft 365 Copilot license, then bills per task.
  • Pay-as-you-go is one cent per Copilot Credit, set by model, context, tools, runtime.
  • Off by default; admins set tenant, group, and user spend caps.
  • It runs on Anthropic Opus 4.8 and Sonnet 4.6 today; GPT-5.5 in preview.
  • An in-house model, Cowork 1, is coming for lower cost.

💬 Smart takes

  • Charles Lamanna, Microsoft EVP: Cowork is the fastest-growing feature in the history of the Frontier program.
  • Skeptic: variable per-task pricing makes budgets hard, and finance teams hate bills they can't predict.

🧭 Where this goes

  1. Likelyusage-based agent billing becomes the enterprise norm by year-end.
  2. Likely'AI credits' shows up as a new line in IT budgets and FinOps tools.
  3. PossibleMicrosoft's Cowork 1 cuts its dependence on Claude within two quarters.
  4. Wild Carda customer gets a surprise five-figure Cowork bill and it goes public.

🥄 The Spoon Take

The seat license is dying for agents. You pay for work done, not chairs filled. Microsoft is also doing something quietly bold: selling an AI coworker on Claude while building a cheaper model to replace it. The runtime owner wins, not the model owner.

🤔 Pushback

Usage-based pricing can stall adoption; if employees fear running up a bill, they just won't use it.

·Microsoft 365 - Cowork is GA·Neowin - Usage-based billing
Monday Jun 15
ENTERPRISEMicrosoft

Microsoft Manages AI Agents Like Staff

15JUN
276,000 ON DUTYAGENT 365

AI agents now get managed like employees. Microsoft's Agent 365 gives every agent an identity, security controls, and monitoring. KPMG just put it across all 276,000 of its people.

The enterprise question flipped. It's no longer 'do these bots work,' but 'who's accountable when they act.'

The platform treats each bot as a tracked corporate identity. Defender logs the devices it runs on, the tools it calls, and the data it can reach. It even spots rivals' helpers like Claude Code.

KPMG switched it on for its whole 276,000-person workforce next to Copilot. Ungoverned AI just became a tracked corporate asset.

full brief & sources

⚡ Why this matters

  • The agent fight is shifting from raw capability to control and governance.
  • Treating agents as identities lets IT manage them with existing security tools.
  • A 276,000-seat rollout signals agent governance is a real budget line, now.

🔍 What happened

  • Agent 365 is generally available; policy controls hit preview in June.
  • Defender maps each agent's devices, MCP servers, identities, and reachable cloud resources.
  • Local agent discovery expands to 18 types, including GitHub Copilot CLI and Claude Code.
  • Pricing: $15 per user per month, or bundled in the new Microsoft 365 E7 SKU.
  • Jun 9: KPMG commits to Agent 365 plus Copilot across 276,000 staff.

💬 Smart takes

  • Microsoft: positions Agent 365 as turning 'shadow AI' into a governed, observable asset.
  • VentureBeat: framed the launch around shadow AI becoming an enterprise threat.
  • Skeptic: governing agents you don't host, including rivals', is far easier to announce than to enforce.

🧭 Where this goes

  1. Likely'agent identity and governance' becomes a standard enterprise procurement checkbox.
  2. LikelySalesforce, Google, and ServiceNow ship competing agent-control planes in 2026.
  3. Possibleagents get HR-style lifecycle management - onboarding, scopes, offboarding.
  4. Wild Carda regulator requires audit logs for autonomous agents acting on customer data.

🥄 The Spoon Take

The model race has a quieter sibling: the control race. Whoever manages, secures, and audits agents owns the enterprise relationship, not whoever has the smartest model. Microsoft is selling agent governance as plumbing. That Agent 365 discovers and manages Claude Code is the whole strategy.

🤔 Pushback

Managing agents you don't run, including competitors' agents, is a bold control claim that enterprises will test hard before trusting it.

·Microsoft - KPMG scales Agent 365·VentureBeat - Agent 365 GA·Microsoft Security - Agent 365 capabilities
Wednesday Jun 10
GOVERNANCEMicrosoft

Microsoft Dev Tools Hacked, Passwords Stolen

10JUN
SUPPLY CHAINGITHUB TOOLSPWNED

The most-used AI coding tools just became an attack surface. Hackers injected malware into Microsoft's GitHub projects used with Claude Code and Gemini CLI. As AI coding agents spread, they become high-value targets for credential theft.

Hackers injected credential-stealing malware into three Microsoft-maintained open-source packages commonly installed alongside Claude Code and Gemini CLI. Scope unknown; Microsoft disclosed on June 8.

The attack vector: a compromised package ran a script harvesting stored API tokens, SSH keys, and browser-cached passwords on install. Rotate all credentials if you used affected versions.

This is a supply chain attack - same pattern as SolarWinds, but targeting AI developer toolchains. A single compromised dev account unlocks both frontier AI APIs and enterprise codebases.

full brief & sources

⚡ Why this matters

  • Every developer building with AI coding tools is running open-source dependencies - this attack shows that dependency chain is now a high-value target
  • A single compromised developer account unlocks both frontier AI API spend and enterprise codebase access - a uniquely valuable attack payload
  • The SolarWinds lesson: sophisticated attackers go upstream in the supply chain, not at the front door - this follows the same playbook

🔍 What happened

  • Microsoft disclosed on June 8 that three GitHub-maintained open-source packages were compromised with credential-harvesting malware
  • The packages are used in toolchains alongside Claude Code and Gemini CLI
  • The malware targeted API tokens, SSH private keys, and browser-cached passwords silently on installation

💬 Smart takes

  • Microsoft detected the breach within 48 hours and yanked affected versions; all users of affected versions should rotate all credentials immediately
  • The security community noted AI developer toolchains are uniquely valuable targets - they bridge frontier AI access and enterprise infrastructure
  • Scope of the breach is unknown - Microsoft has not confirmed how many developers installed the compromised versions

🧭 Where this goes

  1. TechCrunch, Microsoft Security Blog, Ars Technica - June 8 2026

🥄 The Spoon Take

The attack surface isn't the model - it's everything developers touch to build with it. Open-source toolchain security didn't matter much when the stakes were website code. It matters enormously when the same toolchain controls access to frontier AI and enterprise codebases. The SolarWinds lesson is being relearned, one npm package at a time.

🤔 Pushback

The breach was detected fast and scope is unknown - this may be a low-impact incident amplified by AI-news hype.

·TechCrunch - Microsoft GitHub hack·Microsoft Security Blog - Disclosure·Ars Technica - Supply chain attack analysis
Wednesday Jun 3
STRATEGYMicrosoft

Microsoft Builds Its Own Brain

2JUN
97% AIME!MAI-THINK-1NO OPENAI

Mustafa Suleyman's Microsoft AI team ships MAI-Thinking-1, the first in-house reasoning model trained from scratch on commercially licensed data with no distillation from OpenAI's GPT series. It scores 97% on AIME 2025.

Microsoft now has a frontier reasoning model it built itself. 35B active parameters, ~1T total in sparse MoE, 256K context. 97% AIME 2025, 94.5% AIME 2026.

This is the strategic break, not yesterday's Project Polaris coding model. MAI-Thinking-1 is frontier reasoning with no OpenAI lineage. Microsoft Foundry runs it in private preview today. MAI-Code-1-Flash also rolled out to GitHub Copilot users in VS Code - Microsoft says it beats Claude Haiku 4.5 on coding price-performance. Combined with the April amendment ending Azure exclusivity for OpenAI, both sides got what they wanted: Microsoft has its own models, OpenAI has its own clouds.

For PMs, the 'Microsoft = OpenAI reseller' mental model is dead. Expect more Microsoft-native models across products. Copilot pricing pressure goes UP for OpenAI as Microsoft can credibly switch defaults.

full brief & sources

⚡ Why this matters

  • First frontier-grade reasoning model from a Big Three lab that's not OpenAI / Anthropic / Google.
  • 'No OpenAI data inheritance' is the moat - Microsoft owns the full lineage.
  • Validates Mustafa Suleyman's hire (ex-Inflection, March 2024) as Microsoft AI head.

🔍 What happened

  • MAI-Thinking-1: 35B active parameters, ~1T total in sparse MoE architecture, 256K context window.
  • 97.0% on AIME 2025, 94.5% on AIME 2026 (math + multi-step reasoning benchmarks).
  • Trained ground-up on commercially licensed enterprise data.
  • No distillation from third-party models including OpenAI's GPT series.
  • Available in private preview through Microsoft Foundry.
  • MAI-Code-1-Flash: 5B-parameter coding model, rolling out to GitHub Copilot individual users in VS Code.
  • Microsoft claims MAI-Code-1-Flash outperforms Claude Haiku 4.5 in coding price-performance.
  • Announced June 2 at Microsoft Build 2026, Fort Mason, San Francisco.

💬 Smart takes

  • Mustafa Suleyman (Microsoft AI CEO): Microsoft's strategy is to build proprietary AI alongside its OpenAI partnership.
  • Microsoft: MAI-Code-1-Flash was built end-to-end by Microsoft using 'clean and appropriately licensed data.'
  • Skeptic: Microsoft's prior MAI models (MAI-1 in 2024) underwhelmed at launch. Private preview benchmarks need independent verification.

🧭 Where this goes

  1. Within 90 days, MAI-Thinking-1 powers a Copilot Pro tier as an alternative to GPT-5.5.
  2. Microsoft enterprise tiers add a 'MAI-only' SKU with data lineage guarantees.
  3. OpenAI's pricing leverage with Microsoft drops as Microsoft has a credible internal alternative.
  4. EU and government customers prefer MAI for sovereignty and data lineage reasons.

🎯 Implication

  • For PMs: If you ship on Azure AI Foundry, MAI-Thinking-1 is now a default option to test.
  • For execs: Copilot economics shift. The 'OpenAI tax' inside Microsoft products now has an internal alternative.
·Microsoft - MAI-Code-1-Flash intro·CNBC - Lessen reliance on OpenAI·TechTimes - First in-house reasoning
Tuesday Jun 2
STRATEGYMicrosoft

Microsoft Cuts The OpenAI Cord

2JUN
INDEPENDENCE DAY OPENAI MICROSOFT

At Microsoft Build today in San Francisco, Satya Nadella unveiled Project Polaris - Microsoft's own AI coding model. It replaces GPT-4 Turbo as the default brain inside GitHub Copilot starting August 2026. The full agent stack ships alongside.

Microsoft just told OpenAI it doesn't need it for the most important AI product Microsoft ships. Polaris is in-house, Windows-aware, and Copilot-default by August.

The full Build keynote was about Windows as an agent platform. Microsoft Agent Framework merges AutoGen and Semantic Kernel into one supported SDK. Computer-using agents in Copilot Studio went GA. Azure Agent Mesh shipped. Copilot Workspace left beta. Project Polaris was the headline, but the read is bigger - Microsoft just declared independence from OpenAI's model roadmap for its largest dev surface.

For PMs at MS partners: assume Polaris becomes the default reasoning engine across Office, Windows, and Azure within 18 months. For execs: the OpenAI-Microsoft exclusivity story is officially over. For developers: GitHub Copilot is now a Microsoft-stack product, not a wrapper around someone else's model.

full brief & sources

⚡ Why this matters

  • First-party Microsoft model dethrones GPT-4 in GitHub Copilot - Microsoft's biggest AI consumer surface.
  • Windows officially becomes an agent OS - Agent Framework, OpenShell, security primitives ship as a stack, not as features.
  • AutoGen + Semantic Kernel converge into one supported SDK. Two years of forked open-source ends in one commercial product.

🔍 What happened

  • June 2, 2026. Microsoft Build opens at Fort Mason, San Francisco. 2,500 developers on-site.
  • Satya Nadella, Microsoft CEO, opens with: Windows is no longer a platform for human users only - agents are first-class citizens.
  • Project Polaris announced as Microsoft's in-house AI coding model. Replaces GPT-4 Turbo as default in GitHub Copilot starting August 2026.
  • Microsoft Agent Framework: production release of the merged AutoGen + Semantic Kernel SDK.
  • Computer-using agents in Copilot Studio: now GA. Agents interact with websites and desktop UIs directly.
  • Azure Agent Mesh: announced as the orchestration layer for multi-agent enterprise systems.
  • Copilot Workspace: out of beta, generally available.
  • Windows Agent Framework: open-sourced.

💬 Smart takes

  • Satya Nadella (Microsoft CEO): Windows is no longer a platform for human users only. Agents are first-class citizens in the runtime, the tooling, and the distribution model.
  • Notebookcheck: The June 2 keynote is the platform-shift moment - Copilot Workspace out of beta, Azure AI Foundry, Windows local AI all in one breath.
  • Windows News: The agenda was Copilot agents dominate, Windows 12 nowhere in sight.
  • Skeptic: Polaris replacing GPT-4 by August is aggressive. If Polaris underperforms in real Copilot workloads, the OpenAI fallback becomes a procurement-line shame for Microsoft.

🧭 Where this goes

  1. OpenAI loses ~30% of Microsoft inference spend by Q1 2027 as Polaris rolls out across Copilot surfaces.
  2. Polaris benchmarks vs Claude Sonnet 4.6 and GPT-5.5 leak within 60 days. If close, Microsoft starts winning bake-offs.
  3. Anthropic and Google rush their own "agent OS" framings - the platform layer is now the battleground.
  4. Computer-using agents on Windows become a procurement line for enterprise IT - asset-management, helpdesk, software testing all rewrite.
  5. Azure Agent Mesh forces an answer from AWS Bedrock and Google Vertex AI within Q3.

🎯 Implication

  • For PMs: if your product runs inside Windows, plan an agent-driven UI test pass by Q4. Computer Use will find your bad UX before customers do.
  • For execs: the Microsoft-OpenAI alliance is now a procurement question, not a partnership. Treat them as separate vendors.
  • For developers: AutoGen and Semantic Kernel users - migrate to Microsoft Agent Framework now. The fork is over.
·ChatForest - Polaris cuts the cord·Windows News - Build 2026 agenda·Technobezz - Agent mode default·Notebookcheck - Build 2026 preview
Monday Jun 1
STRATEGYMicrosoft

Copilot Hits The Meter

30MAY
$29 to $750 WALLET METERED

GitHub Copilot moves to usage-based billing on June 1. Same monthly fee now buys a fixed AI Credit allowance. Overruns cost real money. Developers posted bills jumping from $29 to $750 and $50 to $3,000 on Reddit and X.

Microsoft just told vibe-coders the party is over. The flat-fee era for premium AI coding ends today.

Each plan still has a flat monthly price. But that price now buys a fixed pot of AI Credits at the model's API rate. Code completions stay free. Agentic sessions burn through credits in minutes. The community thread has 400+ comments and 900+ downvotes. The math is simple: Microsoft was subsidizing a lot of compute, and they're done.

For PMs at AI SaaS vendors: your flat-fee AI tier is on borrowed time. For devs: check the per-token API rate of the model your agent calls before you start a session. For execs: this is the start of repricing across the developer-AI category. Cursor, Replit, and Codex CLI will follow within the quarter.

full brief & sources

⚡ Why this matters

  • First major dev-AI tool to abandon flat-fee pricing for agentic workloads. Sets the reference point everyone else will be compared to.
  • Tells you the unit economics of subscription AI coding never worked at high usage. The honest pricing for power users is per-token.
  • Speeds up consolidation: Cursor, Replit, and Codex have to choose between subsidizing power users or repricing.

🔍 What happened

  • June 1, 2026. GitHub Copilot moves all plans to usage-based billing. Effective today.
  • Pricing: Pro stays $10/month, Pro+ $39/month, Business $19/seat, Enterprise $39/seat. Each monthly fee buys a fixed AI Credit allowance (1 credit = $0.01).
  • Code completions and Next Edit suggestions stay included. Agentic sessions, chat with premium models, and multi-step flows burn credits at the model's API rate.
  • Token usage covers input, output, and cached tokens at posted API rates.
  • Community thread (discussion #192948): 400+ comments, ~900 downvotes within days.
  • TechCrunch May 30 cites Reddit examples: $29/month going to $750, $50/month going to $3,000.

💬 Smart takes

  • GitHub blog: usage-based billing aligns pricing with the cost of running the underlying models - the official framing.
  • Reddit dev: "What a joke. This new usage model is just stupidly expensive. I'm adjusting mine by cancelling."
  • Reddit defender: "The only way it gets crazy like that is if you are purely vibe coding with bloated iterations. It's pretty affordable for even small outfits if used as a tool."
  • Skeptic: Microsoft trained users to lean on Copilot for everything, then changed the contract. The trust hit lands hardest on individual devs and small teams.

🧭 Where this goes

  1. Cursor announces a credit-style pricing tier within 60 days, or holds and bleeds free agentic users.
  2. Codex CLI and Claude Code stay flat-fee through Q3 to win developer market share.
  3. Vibe-coding category compresses. Tools that subsidize unlimited agentic runs become a temporary phenomenon.
  4. Enterprise wins out: token budgets at the seat level become a procurement line, not a developer experience question.
  5. A wave of open-source local-model setups (DeepSeek, Qwen) gets adoption from devs who refuse to meter.

🎯 Implication

  • For PMs: if your AI feature is metered at the model API rate, your customers will start asking for credit caps. Build them.
  • For devs: learn the token math. Premium model with agent loop = $30-$40 per session. Plan accordingly.
  • For execs: dev-AI spend just became a top-line cost center. Budget for it in FY27 the way you budget for cloud.
·TechCrunch - Devs slam token billing·GitHub Blog - Usage-based billing·GitHub - Discussion #192948·GitHub Docs - Models and pricing
Sunday May 31
STRATEGYMicrosoft

Copilot Becomes The Default

28MAY
OFFICE 365FREE BUNDLE

Microsoft announced on May 28 that Business Standard and Business Premium SKUs will include Copilot at no extra cost starting July 1, 2026. The standalone $30/seat Copilot add-on goes away for SMBs (1-300 seats). Standard runs $23.50/user/mo, Premium $32/user/mo, both annual.

Microsoft just stopped selling Copilot as an add-on. Below 300 seats, it's the product.

The unbundled $30/seat price had a slow start. Most SMBs didn't add it. Microsoft's move folds AI into the SKU customers already buy. Productivity apps + AI + connectors + security in one bill. Google's response will land within the quarter - Workspace can't keep AI premium-priced once Microsoft bundles. The pricing floor for productivity AI just collapsed at the SMB tier.

For PMs: stop treating AI features as upsell tier. Your competitors will bundle. For execs: renegotiate your seat economics before competitors do. For investors: Microsoft just compressed AI ARPU at the SMB tier. Q3 will show what that does to per-seat margins.

full brief & sources

⚡ Why this matters

  • First time a hyperscaler folds frontier AI into the base productivity SKU instead of pricing it as a premium add-on.
  • Sets the SMB pricing reference point. Google Workspace and Apple iWork will be measured against this within 90 days.
  • Tells you where the AI-feature business model is going: table stakes, not premium tier.

🔍 What happened

  • May 28, 2026. Microsoft 365 blog announces Business Standard with Copilot and Business Premium with Copilot launch July 1.
  • Business Standard with Copilot: $23.50/user/month annual. Includes Office desktop + web apps, Copilot, connectors, Outlook, OneDrive 1TB.
  • Business Premium with Copilot: $32/user/month annual. Adds Intune device management, advanced security, Defender for Business.
  • Seat cap: 1-300 employees. Above 300, customers stay on Enterprise SKUs where Copilot remains a separate $30 add-on.
  • Same announcement introduces a redesigned Microsoft 365 Copilot UI with Word, Excel, PowerPoint, and Outlook integration.
  • Replaces the previous structure where Copilot was a $30/seat add-on to any base SKU.

💬 Smart takes

  • Microsoft 365 blog (May 28): "world-class productivity apps, AI that's built for work, and the security to help protect employees, data, and IP" - the new pitch to SMBs.
  • Partner channel framing: the bundle is positioned as "the new standard for small business" - Microsoft is explicitly defining AI as table stakes, not premium.
  • Skeptic: Microsoft tried this exact playbook with Teams in 2017 and got hit with EU antitrust. Bundling Copilot into the base SKU at SMB tier may invite the same scrutiny in Brussels.

🧭 Where this goes

  1. Google announces Workspace + Gemini bundled at a similar price point within 90 days.
  2. Salesforce, ServiceNow, and Workday reprice their per-seat AI agents downward in next earnings cycle.
  3. Standalone Copilot revenue line in MSFT earnings disappears or gets reclassified by FY27 Q2.
  4. EU and UK CMA open preliminary inquiries on AI-bundling-into-base-SKU by end of 2026.

🎯 Implication

  • For PMs at SaaS vendors: if your AI feature is a $X/seat upsell, your model is on borrowed time. Re-architect for bundle or pay-per-action.
  • For procurement teams: ask Microsoft, Google, and Salesforce reps for the bundled price BEFORE the add-on quote. The price you get next quarter will be different.
  • For execs: AI inside the seat is now the default expectation. The premium tier needs to deliver agent autonomy or domain depth - not just "AI features."
·Microsoft 365 blog - SMB bundle launch·Microsoft 365 blog - Copilot redesign·Microsoft Partner Hub - SMB momentum·Windows Forum - July 1 launch
Thursday May 14
RESEARCHMicrosoft

Patch Tuesday Just Got Heavier

12MAY

Microsoft ships MDASH, a security system built from 100+ specialized AI agents (auditor, debater, prover) working as an ensemble. Found 16 new Windows vulnerabilities in one Patch Tuesday.

Single-agent thinking is over. Multi-model ensemble with role-specialized sub-agents is the new reference architecture.

The harness does the work. The model is one input. Auditor plus debater plus prover splits beat any single-model prompt-loop on hard problems. The pattern transfers from security to legal review, medical chart review, financial audit.

Expect XBOW, ZeroPath, and ProjectDiscovery to get acquired within 12 months. For PMs building high-stakes agentic products: stop optimizing prompt loops. Design the harness.

full brief & sources

⚡ Why this matters

  • Single-agent thinking is over.
  • Multi-model agentic ensembles are the new reference architecture for high-stakes products.
  • Microsoft just shipped the pattern at scale.

🔍 What happened

  • May 12, 2026. Microsoft ships MDASH multi-model agentic scanning harness.
  • Taesoo Kim (VP, Agentic Security) + Team Atlanta DARPA AIxCC pedigree.
  • 100+ specialized agents (auditor, debater, prover) across frontier + distilled + counterpoint ensemble.
  • 16 net-new Windows CVEs in May 12 Patch Tuesday.
  • 4 Critical pre-auth RCEs (tcpip.sys SSRR UAF, ikeext.dll IKEv2 double-free, netlogon CLDAP, dnsapi UDP DNS heap OOB).
  • 21/21 on StorageDrive private bench with 0 false positives.
  • 96% recall on 5-year CLFS MSRC backlog. 100% on tcpip.sys.
  • 88.45% CyberGym leaderboard top, ~5 points ahead.

💬 Smart takes

  • Taesoo Kim: "The harness does the work, and the model is one input. Single-model harnesses undersold what models can do. Over-trusted single agents overshoot."
  • Kim's framing: "Not which model does it use, but what does it do with the model, and what survives when the next model arrives."
  • Skeptic: 16 CVEs is impressive, but MSRC normally ships 80-150 per Patch Tuesday. MDASH is a meaningful input, not the autonomous SOC story. 88.45% CyberGym lead over 83.1% next entry is smaller than absolute numbers suggest.

🧭 Where this goes

  1. "Multi-agent ensemble" becomes the standard pitch for high-stakes agentic products by Q3.
  2. Google, Anthropic, CrowdStrike, Palo Alto, Wiz ship competing ensemble harnesses within 6 months.
  3. Bug-bounty economics compress hard for routine vulnerability classes.
  4. First publicly-attributed agentic-system exploitation campaign lands within 12 months.

🎯 Implication

  • For PMs building high-stakes agentic products: stop optimizing prompt-and-tool loops. Design the ensemble harness instead.
  • Three concrete moves: split agent into role-specialized sub-agents (auditor, debater, prover); introduce adversarial-debate stages; build plugin extensibility for domain-specific context.
  • For security and platform leaders: sign up for MDASH private preview. Architecture is more transferable than specific numbers.
·Microsoft Security Blog - Defense at AI speed: MDASH multi-model agentic security system (Taesoo Kim, May 12)·Microsoft - MDASH private preview signup