Microsoft is moving 6,000 engineers into your building
The news: Microsoft launched Frontier Company on July 2, a $2.5 billion unit that embeds 6,000 engineers inside customer organizations to ship AI. Amazon committed $1 billion to a similar group on June 30, and OpenAI and Anthropic built their own in May.
Why it matters: This is vendors conceding that pilots stall without hands inside your walls. Embedded engineers move AI to production faster, but they also learn your architecture, your data, and your workflows from the inside. That knowledge becomes leverage at renewal, and it deepens lock-in in ways a license never could.
What to do: Before you accept embedded staff, get it in writing: who owns the resulting IP and models, and what the exit terms are.
A CVSS-10 flaw hands attackers every endpoint your MSP manages
SimpleHelp's remote-management tool has a perfect 10.0 authentication-bypass flaw, CVE-2026-48558. Attackers are exploiting it now. A forged login token creates an admin account, which cascades into every endpoint the provider touches. CISA added it to the KEV catalog on June 29 with a three-day federal deadline. Most enterprises do not run SimpleHelp, but their MSP might. Ask your provider this week whether they patched it, and get the answer in writing.
Meta wants to be your next cloud landlord
Meta confirmed on July 1 that it is building a business to sell its excess AI compute. It joins a market where three hyperscalers still control most capacity, and supply stays tight into 2027. A fourth serious seller gives buyers rare leverage on price and terms. If you are negotiating a compute commitment this quarter, name Meta as an alternative. See if your incumbent sharpens the quote.
Watch This
The forward-deployed engineer wave is quietly moving AI off your software line and onto your services line. Microsoft, Amazon, and others are embedding staff to run deployments. The cost shifts from a license you renew to a services engagement you staff around. Expect Q3 budget reviews to surface a new question: are you buying a tool you own, or renting a team you cannot easily replace?
This week, DoGood network members are pressing vendors on who owns the AI outcome before they let embedded engineers inside. If you run IT or security at a $100M+ company, that ownership question is the one your peers are negotiating right now.
Know a CIO who needs this? Forward it and they can subscribe here.
Enterprise IT leader at a $100M+ company? Apply to join DoGood.