Build1 distinct publisher3 min readUpdated
According to the Pragmatic Engineer, Meta has started making one-off equity counteroffers it never used to make, and UK layoff notices were reversed after staff had already begun interviewing.
The Engineer · Build desk
Compiled by The EngineerSomething wrong?How this is made
Meta cut 10% of staff while revenue and profits were at an all-time high, then reassigned roughly 20% to 30% of its software engineers to data labeling with basically no notice [1][2]. According to Gergely Orosz of the Pragmatic Engineer, the bill is now arriving in the form of senior resignations and one-off equity grants that Meta, by the account of long-tenured engineers he spoke to, previously refused to make at all [3][8].
Start with the mechanic that is easiest to underestimate. In the UK, mass layoffs require a notification to those potentially affected before the cuts can happen [5]. Orosz says he confirmed that a few weeks after that notification, a good chunk of people were "un-notified" [6]. But the people on notice had already started looking for jobs [7]. The notification created a search cohort; withdrawing it did not un-create one. Anyone running a reorg in a jurisdiction with statutory notice should read that as a fixed cost of even a reversed decision.
The counteroffer data is the clearest signal. Retainer equity is now going to IC6-and-above engineers who resign for Google, Anthropic and OpenAI, a practice Orosz says was not done before [9]. He talked with seven such people, all IC6 (staff) or IC7 (principal), and is not aware of IC4 or IC5 engineers getting the same [10][11]. Grants run from $400K to $1M-plus, vesting over three years [12]. The $1M-plus figures all went to engineers holding Anthropic or OpenAI offers; Orosz confirmed a $400K and a $600K grant for engineers with offers from smaller AI startups [13][14]. Annualised, that is roughly $133K to $333K per year of extra equity per retained senior engineer [1]. No offer letter was required: of two recipients he asked, neither had to produce one, and a director together with HR made a discretionary grant once the resignation was tendered [15].
One case, confirmed by Orosz, shows why paying at the exit is bad procurement. A senior, long-tenured engineer was force-reassigned to an AI data labeling team, decided that was not the job they wanted, and interviewed out [16]. Google's offer came in below their current compensation, and they decided to leave anyway [17]. On hearing the resignation, their new manager came back with a large one-off retainer grant vesting over three years [18]. The engineer took Meta's number to Google as evidence they were a key hire, Google raised its offer above the retainer, and the engineer left happily [19].
The retainers are also not converting. Orosz confirmed three engineers with Anthropic offers were countered with $1M-plus grants vesting over four years; two rejected and joined Anthropic, and the third accepted and then left a month later anyway, forfeiting the grant [20][21][22]. That is zero of three retained on that sample [2]. He reports OpenAI is having similar success, and that the two labs are the main destination for ex-Meta engineers, since both can match Meta's total compensation and both organise secondary equity sales despite not being publicly traded [23][24][25].
Worth watching: whether the retainers move below IC6, whether the un-notified UK cohort stays, and whether Meta's data labeling reassignment survives contact with the attrition it caused. All of the above is single-sourced to Orosz's reporting and his confirmations with named-level engineers, not to Meta [3].
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
Meta did 10% layoffs at a time when its revenue and profits hit an all-time high.
In the UK, mass layoffs require a notification to those potentially affected before the cuts can happen.
A few weeks after the UK notification, a good chunk of the notified people were 'un-notified', which Orosz says he confirmed.
Those who had been put on notice had already started to look for jobs.
Long-time Meta engineers told Orosz that Meta simply did not make counteroffers or negotiate when an engineer handed in a resignation, and that this practice has been abandoned.
Meta started offering large retainer equity grants to IC6-and-above engineers resigning to join Google, Anthropic and OpenAI, a practice not done before.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single-source, anonymous, self-confirmed
Every factual element traces to one newsletter issue by one reporter relying on unnamed current and former Meta employees, with 'I confirmed' as the verification standard. No documents, offer letters, company statement, or corroborating publisher appears in the cluster. Specific small-sample details (seven retainer recipients, three Anthropic cases, two counteroffer recipients asked about offer letters) are internally consistent and unusually concrete, which lifts this above rumor, but the broad quantities (20-30% reassignment, 'a good chunk' interviewing) are unverifiable as stated.
Confirmed in a small sample at one employer
The retainer practice is documented as actually happening rather than announced: seven recipients described terms, two described the trigger mechanics, one full Google outbid case is narrated, and three Anthropic counteroffers with outcomes are reported. But the practice is confined to one company, one seniority band, and a recent few-week window, with no count of total grants, no share of resigners covered, and no evidence it exists outside Meta. UK notice reversals are likewise reported only qualitatively.
Specifics oversold as pattern
The reporting itself is hedged ('seems to be', 'I confirmed'), but the framing generalizes a handful of confirmed cases into a company-wide practice shift and a 'bleeding talent' verdict that the disclosed sample cannot carry. Headline-grade figures like '$400K-plus retainers' come from a self-selected group of seven who chose to talk to a reporter, and the two-lab destination claim is contradicted in part by the same issue's Google and startup examples plus one engineer who stayed. The gap is moderate rather than severe because the mechanics and amounts are reported precisely and the author flags his own prior position.
Subscription funnel plus prior-position defense
The item is explicitly a free 'bonus' extract of a paid issue, opening with a subscribe pitch and noting full subscribers got it seven days earlier, so scoop-grade Meta attrition detail directly serves conversion. The author also restates and defends his own two-month-old thesis that the layoffs were self-inflicted, giving a consistency incentive to favor confirming anecdotes. Sources are anonymous engineers who benefit from a narrative that senior Meta talent is scarce and expensive to keep. None of this is disqualifying, but it shapes selection.
Plausible and specific, but unconfirmed
Confidence is limited by structural factors rather than internal contradiction: one publisher, anonymous sourcing, no company response, and no way to size the practice. The granular, mutually consistent mechanics (level bands, vesting schedules, HR-plus-director trigger, named destination employers) and the author's track record of covering this beat support the narrow case-level claims; the company-scale generalizations should be held loosely pending corroboration.
product
Washington's secret AI test is coming for open weights, and release dates go with it2 distinct publishers
security
The nationalization argument is really a vendor-continuity memo1 distinct publisher
invest
Google Ships Flash Instead of Pro While OpenAI Loses Its Two Best Operators1 distinct publisher
invest
The labs got better at watching their agents escape. They did not get better at stopping them.1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 20, 2026