Petronella.ai

Why the Engineering Manager Role Is Under Attack Again

November 22, 2026 · Cybersecurity
Why the Engineering Manager Role Is Under Attack Again

The Engineering Manager role is being cut again across Big Tech, and the flattening cycle now underway carries real operational and security risk for regulated businesses.

This analysis draws on a recent article published at manager.dev, which examines why companies are cutting managers and what happens next. The article builds on a conversation with Suzan Bond, an organizational development consultant and former COO who writes at Field Notes and runs Constellary. Reporting from Business Insider documents the Microsoft cuts in detail. If you lead engineering, security, or compliance at a regulated company, this cycle deserves your attention before it reaches your org chart.

Key Takeaways

What Happened: The Facts in Detail

The current wave of manager cuts

The manager.dev article describes a pattern: larger technology companies began laying off middle managers, and smaller companies started to consider doing the same, whether or not their own conditions matched. The stated rationale is familiar. With AI, leaders believe they can run leaner, and agents can replace much of the human work. The first target of this claim was software engineers. For a short while, leaders declared the profession dead, but that reversed quickly as companies realized they still need experienced engineers to decide what to build, review what large language models produce, and maintain systems afterward. Engineers now make up 55% of Big Tech hiring, up from 46% in 2019.

Managers are harder to justify in that framing, so the flattening cycle returned. When leaders look at a first-line Engineering Manager, the thinking goes: they are close to the work, so maybe they should just do the work, maybe fewer of them are needed, and what value do they actually provide?

The numbers behind the flattening

Business Insider reported in April 2025 that Microsoft was cutting thousands of jobs to increase what it calls "span of control." Microsoft confirmed the cuts on a Tuesday, saying about 3% of its global workforce, roughly 6,000 employees, would be affected. Affected US staff stay on the payroll for 60 days, though that varies globally based on local regulations.

Microsoft has no centralized span-of-control goals and no stated maximum for reports per manager, but individual leaders set their own targets. Cloud and AI boss Scott Guthrie wants eight engineers per engineering manager and seven project managers per manager. Microsoft Security chief Charlie Bell is going for nine reports per manager, without separating engineers from project managers. A Microsoft spokesperson said there are no blanket ratio numbers.

The cuts also target non-coders to increase the number of coders on projects, an internal effort Microsoft calls its "PM ratio." The context matters: Microsoft is reducing costs amid significant investment in AI, and Barclays analysts called the cuts "a commitment to profitable growth and optimizing the company's workforce, particularly with AI becoming a more prominent and efficient tool within the organization."

Google's Sundar Pichai told staff in December that the company cut vice president and manager roles by 10% as part of an efficiency drive. Amazon has been trying to increase the ratio of individual contributors to managers. Meta, per the manager.dev article, has stated a goal of one manager to 50 engineers, and 15 to 20 engineers reporting to a manager is described as very common. Meta also laid off 700 engineering managers.

Why insiders at Microsoft mostly approved

A half-dozen Microsoft insiders who spoke to Business Insider generally considered the flattening a good thing. One executive said there are a lot of managers who just are not that great, and another said the company has too many levels and layers. The layers accumulated because Microsoft often rewarded good engineers by promoting them into management, sometimes with as few as one or two reports, and those engineers-turned-managers often still spent most of their time in the codebase and were not very effective as managers. Still, insiders expressed uncertainty about how Microsoft will collapse its management layers and make the result work, and one staffer said managers are particularly concerned.

This has happened before, and it failed

The manager.dev article traces the history. In 2002, when Google had roughly 300 engineers, Larry Page and Sergey Brin tried to remove all managers entirely. It failed within a couple of months. They relented when too many people went directly to Page with questions about expense reports, interpersonal conflicts, and other nitty-gritty issues. As the company grew, the founders realized managers contributed in other important ways: communicating strategy, helping employees prioritize projects, facilitating collaboration, supporting career development, and ensuring processes and systems aligned with company goals.

Six years later, with a workforce grown to tens of thousands of employees, Google ran Project Oxygen, which showed that managers did matter, with good managers strongly correlating with team performance and engineer satisfaction. Google is now flattening again.

Microsoft has its own history here. In 2014, Satya Nadella announced a major restructuring with fewer management layers and wider spans of control. Eleven years later, in 2025, Microsoft flattened again to increase span of control.

The Engineering Manager Role: How the Erosion Works and Why It Matters

The core insight from the manager.dev article is that the Engineering Manager role does not usually get eliminated in one decision. It erodes through a self-reinforcing cycle. When a manager's span of control expands and they are also expected to ship code, capacity for actual management shrinks. With less capacity, they do less of the work only they can do: fewer one-on-ones, postponed career conversations, unresolved team conflicts and process problems. Providing less value, they matter less to the engineers reporting to them, and those engineers get used to managing without them. The less value the leader perceives, the more the engineer-to-manager ratio increases again, restarting the cycle.

The article argues the underlying failure mode is organizational debt, a gradual buildup of dysfunction that undermines how good work gets done, analogous to technical debt. Relational debt accumulates fastest after a flattening. It describes strained, neglected, or transactional relationships where trust and psychological safety erode. When an Engineering Manager's role expands to be both more technical and more heavily loaded with reports, something has to give, and what gives is the people part of the company.

The visible symptoms are more negative behavior turning into toxic behavior, more conflicts, and less collaboration. There is a compounding effect: engineers are also less in contact with each other, and managers report being asked how to keep teams together when team members prefer talking to AI rather than to each other. When conflict surfaces, leaders often treat it as a single bad apple and remove the problematic person. The article's counterpoint is that the person operates inside an environment, and a poor environment does more damage to company goals than shipping speed ever makes up for.

Why does the pendulum keep swinging back? The article's answer is that companies can run flat for a while. It takes months or years before the impact shows up, and no company suddenly hires hundreds of managers. Recovery is slow and quiet: layers get rebuilt over time. The question for any leader watching Big Tech headlines is whether their own conditions actually match, or whether they are importing a delusion.

For businesses in regulated industries, there is an additional layer. Modern security frameworks increasingly treat governance and human accountability as first-class requirements. CISA's Cross-Sector Cybersecurity Performance Goals 2.0, updated to align with NIST CSF 2.0, added a GOVERN function that regroups goals around leadership accountability, oversight, and risk management. NIST SP 800-61 Rev. 3, the incident response guidance finalized in April 2025, ties incident response into organization-wide cybersecurity risk management under the CSF 2.0 functions. Those functions are executed by people who talk to each other, escalate problems early, and trust their counterparts. Strip out the management layer that maintains those relationships and you have not just an organizational problem. You have thinned the human fabric that detection, escalation, and coordinated response depend on. The AI-Driven SOC: Autonomous Threat Response model works precisely because humans remain in the loop; the same principle applies inside your engineering organization.

What the Evidence Does and Does Not Show

The sources are candid about their limits, and so should you be.

First, much of the Microsoft reporting rests on anonymous insiders. Business Insider notes that a half-dozen Microsoft insiders spoke on condition of anonymity, and that Microsoft's spokesperson disputed the existence of blanket ratio numbers even while individual leaders' targets were described. The characterization of why Microsoft accumulated so many layers comes from those unnamed staffers, not from an audited internal review.

Second, the causal claim connecting Meta's layoff of 700 engineering managers to a subsequent resignation wave comes from the manager.dev article, which asserts the timing but does not establish causation. Resignations have many drivers, and the article does not claim to have isolated this one.

Third, the historical evidence cuts both ways. Google's 2002 experiment and Project Oxygen show that removing managers entirely failed, but they do not prove that every current cut is a mistake. The manager.dev author is explicit about this: more management layers are not always better, and there is real benefit in simplifying the chain of command. The distinction the author draws is between cutting managers because managers supposedly do not matter, and deliberately fixing your own org structure based on your own conditions. Microsoft's insiders, notably, mostly approved of the flattening while remaining uncertain whether it would work.

Fourth, the timeline of consequences is not precisely known. The article says it takes a few months or years before companies feel the impact of running flat, but it does not quantify how much relational debt a given span-of-control increase produces, or at what point recovery becomes necessary. What happens at a one-to-fifty ratio versus one-to-twenty is asserted as a goal and a common pattern, not as a measured outcome study.

Finally, the article's prescriptions, such as writing down the ten to fifteen smaller problems behind a big one and translating them into obstacles to business goals, are advice from practitioners, not validated interventions. They are reasonable, and they are unproven in the way most management advice is.

What remains genuinely unknown is how this cycle resolves at AI-era scale. The prior experiments at Google and Microsoft predate widespread AI coding agents. Whether the current flattening produces the same quiet rebuild, or something structurally different, is a question the sources raise but cannot answer.

The Compliance and Risk Implications for Regulated Organizations

For most businesses, cutting managers is an organizational gamble. For regulated businesses, it is also a compliance problem. The obligations that govern defense contractors, healthcare entities, and payment processors are not just technical controls. They are governance and accountability requirements, and governance is exactly what flattening removes first.

Consider what the NIST Cybersecurity Framework 2.0 did when it was released in February 2024: it added a sixth function, GOVERN, alongside IDENTIFY, PROTECT, DETECT, RESPOND, and RECOVER. CISA's Cross-Sector Cybersecurity Performance Goals 2.0 were updated to align with that structure, and the update explicitly regroups goals around leadership accountability, oversight, and risk management. The federal baseline for critical infrastructure now assumes someone is accountable. When a company removes the layer of people whose job is accountability, it creates a gap that an auditor, an assessor, or an investigator will eventually find.

The same pattern shows up in incident response. NIST SP 800-61 Rev. 3, finalized on 04/03/25, frames incident response as something woven throughout cybersecurity risk management activities, not a runbook pulled off a shelf during a crisis. Preparation, detection, response, and recovery all depend on people who know the systems, know the team, and have the authority to act. An engineering manager with 15 to 20 direct reports, or a Meta-style goal of 1-to-50, cannot have those relationships. As the source article documents, the first things to go in the erosion cycle are 1:1s, career conversations, and conflict resolution. In a regulated environment, those "soft" activities are the same activities that surface an insider threat, a misconfigured system, or an engineer cutting corners on evidence handling before it becomes a reportable incident.

There is also a documentation dimension. Frameworks such as CMMC Level 2, built on the 110 controls of NIST SP 800-171 Rev 2, and the HIPAA Security Rule at 45 CFR Part 164, with its 18 standards across administrative, physical, and technical safeguards, require defined roles and responsibilities. The Petronella vCISO practice designates a named security officer of record for HIPAA (45 CFR 164.308(a)(2)), CMMC senior official authorization, SOC 2 defined roles, and PCI assigned responsibility, precisely because regulators and auditors ask who is responsible, not whether a committee somewhere might be. Flattening does not eliminate that question. It just leaves it unanswered.

Finally, the relational debt described in the source article has a security corollary. Strained, neglected, transactional relationships erode psychological safety, and engineers who stop talking to each other also stop reporting anomalies to each other. The article notes managers asking how to keep teams together when engineers prefer talking to AI rather than to each other. For a security leader, that is not a culture nicety. It is a detection problem.

What This Means for Regulated Industries

Defense Contractors

Defense contractors face the sharpest version of this problem because CMMC accountability is contractual, not aspirational. Under DFARS 252.204-7019, a contractor must maintain a current NIST SP 800-171 self-assessment score posted to the Supplier Performance Risk System, scored on a range from minus 203 to 110 with 1, 3, or 5 point control weights. Those scores depend on control owners who actually know their systems. When a reorg scatters control ownership across overloaded managers, scores drift from reality, and a C3PAO assessment will expose the gap.

The incident reporting duty compounds it. DFARS 252.204-7012 requires reporting a cyber incident affecting covered defense information to DoD within 72 hours of discovery via dibnet.dod.mil, using a DoD-approved medium assurance certificate obtained in advance, followed by preserving affected system images and monitoring data for at least 90 days. That 72-hour clock does not pause while a flattened organization figures out who is in charge. If your incident response plan names roles that no longer exist, you have a plan that fails exactly when you need it. Note also the current regulatory uncertainty: the DoD CIO suspended the 2026-11-10 Phase II deadline on 2026-07-13, with Phase I self-assessments remaining in force and later milestones paused. A pause is not a pardon. Contractors should use the window to fix organizational gaps, not to defer them.

Contractors preparing for CMMC Level 2 should also remember that readiness is a 12 to 24 week initial build followed by years of POA&M tracking and annual evidence refresh. That ongoing work requires stable ownership. A company that reorganizes its engineering management every 18 months will struggle to keep an SSP, a POA&M, and 14 control-family policies current.

Healthcare

Healthcare organizations live with the consequences of thin oversight in a very concrete way. The IBM 2026 figure puts the average healthcare breach at $6.64M, and Change Healthcare's breach ultimately reached 192.7M individuals in the final HHS filing of 2025-07-31. Business associates accounted for about 66% of breached healthcare records in 2024, which means vendor oversight is a first-order risk, and vendor oversight is a management activity, not a tooling activity.

The HIPAA Security Rule requires a Risk Analysis under 45 CFR 164.308(a)(1)(ii)(A) and administrative safeguards that assign responsibility. HHS OCR enforcement shows what happens when accountability lapses: Children's Hospital Colorado received a $548,265 civil money penalty in December 2024 with MFA disabled, and an April 2026 ransomware sweep totaled $1.165M. None of those penalties were caused by a lack of managers per se, but every enforcement action presumes someone was accountable for the control that failed. Flattened organizations should map every HIPAA safeguard to a named individual and verify that person has the capacity to actually exercise the responsibility.

Legal

Law firms and legal technology providers sit in an unusual position: they hold client confidential data, they are often treated as business associates under HIPAA when they handle protected health information in litigation, and they increasingly face client security questionnaires modeled on SOC 2. The AICPA SOC 2 framework covers five Trust Services Criteria, and most SaaS buyers ask for Type I first, then Type II after six to twelve months of evidence. Evidence is the operative word. Type II audits examine whether controls operated effectively over time, and controls operated by overloaded managers who no longer hold 1:1s or review their teams' work produce thin, inconsistent evidence.

The practical guidance for legal-sector IT leaders is to treat any management flattening as a change requiring a control impact assessment. If the person who approved system access changes, or the person who reviewed vendor security, no longer exists as a role, the control narrative in your SOC 2 documentation needs updating before the auditor asks.

Financial Services

Financial services firms operate under PCI DSS v4.0.1, which covers 12 control areas across cardholder data environments, alongside examination and reporting obligations that presume functioning oversight. The industry also faces acute third-party and AI-related risk: the Verizon 2025 DBIR found third-party involvement in breaches rose from 15% to 30%, and IBM's 2025 Cost of a Data Breach report found 20% of breaches involved shadow AI. Both risks grow fastest in organizations where engineers work in isolation, ship quickly, and have no manager with the capacity to ask what they built and where the data went.

For financial-sector leaders, the flattening wave is a reason to audit your own span of control before your board or an examiner does. If your engineering managers carry 15 to 20 reports and are also expected to ship code, your change management, access review, and vendor approval controls are running on borrowed time, whatever your policy documents say.

An Action Plan for Security and IT Leaders This Month

  1. Map every compliance control to a named owner. Pull your CMMC, HIPAA, PCI DSS, or SOC 2 control matrices and confirm each control has a current, named owner. Flag every control still assigned to a role that was eliminated or merged in a recent reorg.
  2. Measure your real span of control. Count direct reports per engineering manager. If you are at 15 to 20, or anywhere near the 1-to-50 goal Meta has stated, document what people-management activities have already been dropped.
  3. Test your incident response plan against the org chart. Run a tabletop exercise and check whether the people your plan names for detection, escalation, and external reporting still hold those jobs. For defense contractors, confirm your dibnet.dod.mil medium assurance certificate is obtainable in advance and that someone can meet the 72-hour reporting requirement.
  4. Write down the small failures. Following the source article's advice, capture the 10 to 15 small problems your teams are accumulating, such as skipped reviews or unresolved conflicts, and translate them into business and compliance risk your executives care about.
  5. Review AI usage on engineering teams. Shadow AI appeared in 20% of breaches in IBM's 2025 report. Ask where engineers are pasting code, data, or client information into AI tools, and whether any manager has capacity to enforce the policy.
  6. Refresh security awareness training. If engineers are talking to AI instead of each other, your human detection layer is weakening. Revisit your security awareness training program with that shift in mind.
  7. Update your documentation of roles. For HIPAA, CMMC senior official authorization, SOC 2 defined roles, and PCI assigned responsibility, ensure formal designation documentation reflects the current organization, not the one from two reorgs ago.
  8. Check your evidence pipeline. If you are in a SOC 2 Type II observation window or a CMMC POA&M cycle, verify that the evidence being collected still matches the controls as they operate today.
  9. Bring the analysis to your board. Frame flattening risk as organizational debt with a compliance cost, not as an argument against efficiency. Boards respond to specific, named gaps better than to general warnings.
  10. Get an outside read on the gaps.

How Petronella Technology Group, Inc. Helps

Petronella Technology Group, Inc. works with regulated businesses and defense contractors whose security programs depend on people, not just tooling. Our compliance practice covers CMMC readiness, vCISO services, HIPAA, SOC 2, and PCI DSS, and every practitioner on the compliance team holds the CMMC-RP designation. The firm is listed on the Cyber AB Marketplace as Registered Provider Organization #1449. A typical CMMC Level 2 readiness engagement runs 12 to 14 weeks for a mid-size contractor, producing an SSP covering all 110 NIST SP 800-171 controls with named owners, 14 control-family policies, a live POA&M, and a mock assessment using the Cyber AB CMMC Assessment Process. We are an RPO, not a C3PAO, and we do not perform certification assessments or promise guaranteed outcomes.

For organizations whose flattened structure has thinned their monitoring and response capacity, our managed detection and response service provides 24/7 threat monitoring, AI-powered analytics, and human-led investigation, replacing the need for an in-house security operations center. The AI-augmented hybrid SOC runs on the company's private AI cluster, which matters for defense industrial base and healthcare clients that cannot send CUI or PHI to a public-cloud SOC. Guardrails are explicit: the AI never closes a ticket on its own, never touches production systems without human authorization, and every action is logged for CMMC and HIPAA audit. MDR reporting is designed to provide the evidence auditors need for HIPAA, CMMC, PCI DSS, SOC 2, and NIST 800-171 verification, which directly addresses the evidence-gap problem a reorg creates. For broader program needs, see our cyber security services and our analysis of AI SOC: XDR, SOAR, and Copilots for Response.

If you need a named security officer of record without a $325K to $450K all-in loaded annual CISO cost, our vCISO tiers provide designated accountability for HIPAA, CMMC, SOC 2, and PCI responsibilities, with a written current-state assessment delivered within 30 days and a first quarterly board memo at day 75 to 90. To understand why clients have trusted the firm since its founding in 2002, see Why Petronella: Founded 2002, BBB A+.

Sources

Frequently Asked Questions

Why are companies cutting engineering managers in 2025 and 2026?

Companies are cutting managers to increase span of control and reduce costs amid heavy AI investment. Microsoft confirmed cuts affecting about 3% of its global workforce, roughly 6,000 employees, explicitly to flatten management layers, and Google's CEO told staff the company cut vice president and manager roles by 10% as part of an efficiency drive.

Does flattening management actually hurt security and compliance?

It can, because accountability and oversight are core to modern frameworks. CISA's Cross-Sector Cybersecurity Performance Goals 2.0 and NIST CSF 2.0 both emphasize the GOVERN function, including leadership accountability and oversight, and incident response guidance in NIST SP 800-61 Rev. 3 assumes prepared people with defined roles. Removing the layer that carries those responsibilities creates gaps auditors and assessors will find.

What span of control is too wide for an engineering manager?

The sources do not establish a definitive threshold, but the trend is striking. Meta has stated a 1-to-50 ratio as its goal, 15 to 20 engineers per manager is becoming common, and Microsoft leaders have set targets such as eight or nine reports per manager. What is known is that the source article documents an erosion cycle: as ratios rise, managers drop 1:1s and conflict resolution, provide less value, and appear less necessary, which justifies raising ratios again.

Have companies tried removing managers before, and what happened?

Yes, repeatedly. In 2002, when Google had roughly 300 engineers, Larry Page and Sergey Brin tried to remove all managers, and it failed within a couple of months as people brought expense reports and interpersonal conflicts directly to Page. Google's later Project Oxygen study found good managers strongly correlated with team performance and engineer satisfaction, yet Google is now flattening again, cutting about 10% of manager, director, and VP roles.

What should a defense contractor do if a reorg eliminated its control owners?

Reassign every NIST SP 800-171 control to a named, current owner and update the SSP and POA&M accordingly, then verify the new owners have capacity to exercise the responsibility. Remember that DFARS 252.204-7019 requires a current self-assessment score in SPRS, and DFARS 252.204-7012 requires incident reporting to DoD within 72 hours of discovery, so the org chart and the incident response plan must match before an incident, not after.

Is the engineering manager role going away?

The historical evidence suggests not. The pendulum has swung before, in 2002 at Google, in 2014 when Satya Nadella announced a major restructuring at Microsoft, and again in 2025 when Microsoft flattened to increase span of control. Companies accumulate organizational debt when they go flat, and they slowly rebuild layers over time. The source article notes that just a couple of months after Meta laid off 700 engineering managers, it faced a serious resignation wave.

If your organization is flattening and you need to know what that means for your compliance posture, call Penny, our AI assistant, at 919-348-4912 to book a consultation, or reach Petronella Technology Group, Inc. through our contact form.

Get the 2026 Cybersecurity Survival Guide

Free, practical, and specific to regulated environments. We will email it to you.

No spam. Unsubscribe anytime.

Talk to Petronella Technology Group, Inc.
Private, on-premises AI and compliance for regulated data. Call 919-348-4912, get a free AI assessment, or explore our AI, cybersecurity, and compliance services.