Is Meta Dismantling Its Own Engineering Excellence?

By Gergely Orosz | The Pragmatic Engineer | June 16, 2026

Leadership at the social media behemoth appears to be on an AI-driven crusade, tearing through the very engineering organization that built its empire.

📢 A Note from Gergely

Hi there! This is a complimentary edition of The Pragmatic Engineer. I specialize in analyzing the frictions and triumphs of Big Tech and startups from the perspective of senior developers and technical leaders.

Pro Tip: Many of my readers use their corporate Learning & Development (L&D) budgets to cover this subscription. If you have one, feel free to ask your manager for reimbursement!

The Sudden Collapse of a Legacy

For two decades, Meta maintained a legendary, high-performance engineering machine. That changed abruptly around April of this year.

For twenty years, the company evolved from a chaotic "break things" mentality to a more mature "stable infrastructure" approach. Engineers were generally empowered, focused on creating tangible impact, and capable of balancing commercial needs with technical integrity.

However, in recent weeks, it feels as though leadership is following a manual on how to destroy a successful engineering culture with maximum efficiency.

📋 What This Analysis Covers

The pre-AI golden era of Meta engineering
The aggressive push toward AI integration
The decline in engineer morale (feeling like "trash")
High-profile outages and internal chaos
The concept of "AI Psychosis"

The Evolution of Meta's Engineering DNA

I categorize Meta's technical history into two distinct epochs:

Era 1: `move-fast-and-break-things`

During the 2010s, Facebook became famous for ignoring industry "best practices" and winning anyway. This culture was codified in a physical, 70-page manual known as the "Little Red Book" (a nod to Chairman Mao's 1964 volume).

The book emphasized speed, ownership, and a total lack of fear. The campus was plastered with mantras such as:

Done is Better Than Perfect
Fail Harder
What Would You Do If You Weren’t Afraid?
Every Day Feels Like a Week
The Wright Brothers Did Not Have Pilot Licenses

🖼️ Cultural Artifacts

The Little Red Book

Era 2: `move-fast-with-stable-infra`

By 2022, the recklessness had subsided. The goal shifted toward maintaining velocity while ensuring the foundation didn't crumble.

As I noted in a previous deepdive:

"The culture is incredibly engineering-centric... This likely stems from Mark Zuckerberg being a coder himself. Impact is the primary currency here."

Comparison of Engineering Eras

Feature	"Break Things" Era	"Stable Infra" Era
Primary Goal	Rapid Growth / Disruption	Scalable Velocity
Risk Tolerance	Extremely High	Calculated
Process	Non-existent	Loose / Minimal
Infrastructure	Experimental	Battle-tested (e.g., Threads launch)
Documentation	Negligible	Still low, but functional

The "Impact" Equation

At Meta, success is measured by individual impact. In mathematical terms, engineers often view their value as: $\text{Individual Impact} = \sum (\text{Measurable Wins}) \times \text{Scale}$ This focus often leads to a preference for short-term, quantifiable victories over long-term collaborative efforts.

The Current Crisis: From Profit to Cost

The current state of Meta can be visualized as a sharp pivot in how the company views its human capital:

The "AI Psychosis"

Leadership is now pressing engineers to integrate AI into everything, regardless of whether it makes sense. This has led to several self-inflicted wounds:

Devaluation of Talent: Core engineers report feeling treated like disposable assets rather than the architects of the company's success.
Technical Debt: The rush to AI has led to some of the most embarrassing outages in the company's history.
Cultural Erosion: The shift from ~~Profit Center~~ $\rightarrow$ Cost Center has happened in a matter of weeks.

💻 The "New" Engineering Logic

If the old culture was a script of empowerment, the new one looks more like this:

def engineering_value(employee):
    if employee.uses_ai_everywhere == True:
        return "Retain (for now)"
    else:
        return "Cost Center / Redundant"

Final Thoughts

Is this "AI Psychosis" unique to Meta, or is it a harbinger for the rest of Silicon Valley? When a founder-CEO who once coded the first version of the product begins to treat his engineering org as a burden rather than a superpower, the results are usually catastrophic.