Meta record profits, low morale: who pays for the AI buildout

Meta has cut equity awards for most of its workforce two years running, suspended share buybacks for the longest stretch since its repurchase program launched in 2017, and issued $55 billion in new debt over the past six months. Analysts expect the company to burn cash in the second half of this year. The earnings headlines remain strong. That gap is what the numbers actually show about how Meta is financing its AI ambitions and who is absorbing the cost.

This piece is about compensation structure and cash flow. Direct data on employee sentiment at Meta is not available in current reporting, but the financial trade-offs workers are navigating are concrete enough to stand on their own terms.

Two straight years of equity cuts

At Meta, equity is not a supplementary perk. For engineers, product managers, and most senior technical staff, stock awards form a core part of total compensation. The company cut annual equity distributions by roughly 10% for most employees last year, a reduction that shocked some staff at the time. This year brought another 5% reduction, affecting tens of thousands of employees, per the same reporting. Compounded across two years, that is a meaningful reduction in real pay even where base salaries have held flat.

Both rounds of cuts were tied directly to Zuckerberg's push to hire top AI researchers and expand Meta's data center capacity, the FT Chinese edition reported earlier this year. The pressure has since deepened: Meta cited rising input costs as it added $10 billion to its capital expenditure forecast, according to the FT. Part of what the broader workforce is absorbing is cost inflation baked into the capital budget itself, not just the price of specific product bets.

The timing matters. These reductions are happening alongside strong stock performance and consistently bullish public communications from Zuckerberg. The employees absorbing the cuts are not sharing in a downturn. The savings are not flowing to shareholders either buybacks have been paused entirely, the longest halt since the repurchase program began in 2017, according to the FT. The money is going into infrastructure whose returns remain uncertain.

Meta employee morale and what the compensation structure actually signals

No current reporting includes internal surveys, retention figures, or named executives discussing staff sentiment at Meta. What it does include is a precise picture of the compensation conditions workers are operating within, and the context around those conditions carries its own weight.

Employees who have seen their equity trimmed two years running can observe that neither shareholders nor the balance sheet are being spared: buybacks paused, debt issued at scale, capex forecast revised upward. The question they are implicitly being asked to answer is whether the AI buildout will eventually justify what they gave up. Zuckerberg's own answer, when pressed by an analyst last week, was that Meta does not have "a very precise plan for exactly how each product is going to scale month-over-month," the FT reported. That is a defensible position for a fast-moving technology bet. It is also the answer tens of thousands of employees are effectively receiving when they ask what they traded their equity to build.

The structural conditions are clear. Whether they register as measurable dissatisfaction at scale is a separate question the current reporting cannot answer.

Why profits look strong while cash runs short

Headline profit and free cash flow are not the same number, and right now at Meta they are pointing in different directions.

Free cash flow is what remains after operating costs and capital expenditure the figure that actually funds buybacks, debt service, and equity awards. Meta is expected to burn cash in the second half of this year, according to Visible Alpha analyst estimates cited by the FT. Strong operating profit can coexist with that kind of pressure when capital spending is aggressive enough.

The sector-wide picture is starker. The combined free cash flow of the four major hyperscalers Amazon, Alphabet, Microsoft, and Meta is projected to fall to roughly $4 billion in Q3 this year, against a post-pandemic quarterly average of approximately $45 billion. Full-year free cash flow across the group is on track for its lowest level since 2014, when these companies' combined revenues were roughly one-seventh their current size, according to Visible Alpha estimates. That gap between revenue scale and cash generation is the most direct measure of what the AI buildout is consuming.

The reported figures may even understate the pressure. Because "free cash flow" has no standardized accounting definition, companies have discretion in how they treat items like stock-based compensation and data center leases. Christian Leuz, an accounting professor at the University of Chicago's Booth School of Business, told the FT that actual free cash flows at major hyperscalers are probably worse than their reported figures. Tech groups including Meta have also shifted tens of billions in data center projects off their balance sheets through special-purpose holding companies, structures that can obscure who bears the risk if AI demand falls short.

The strategic logic, and what employees are being asked to trust

The competitive rationale for spending at this scale is not irrational. Bank of America analyst Justin Post described Big Tech's collective $725 billion AI commitment as the deepest industry-wide capital expenditure cycle the sector has seen, calling it a "once in a lifetime opportunity." Leuz framed the dynamic more structurally: companies feel compelled to invest when rivals are investing, making it a prisoner's dilemma that reinforces the spending cycle regardless of whether individual bets prove out, per the FT.

That logic does not make the payoff predictable. Leuz also noted that heavy fixed-asset investment cycles in industries like telecoms and chemicals have historically produced overcapacity, compressed margins, and weak returns when infrastructure buildout outpaces demand, per the FT. The arms-race framing executives use to justify the spending is precisely the dynamic that has generated excess in previous cycles.

Post offered a counterpoint worth taking seriously: the major hyperscalers entered this capex cycle with strong balance sheets, which reduces the risk of adding debt during a short period of negative free cash flow, per the FT. Amazon's Andy Jassy made a similar case to investors, comparing the AI buildout to the early years of AWS, which weighed on Amazon's balance sheet for years before becoming the source of more than half its profits, per the FT. Whether Meta's infrastructure bets follow a comparable arc is, in effect, what tens of thousands of employees are being asked to bet their compensation on.

What to watch in upcoming earnings

Analysts project cash generation will improve across the hyperscalers next year as AI infrastructure begins producing revenue, per the FT. Whether that holds at Meta will not be visible in headline profit figures. Four indicators will show it first: whether free cash flow recovers as projected, whether debt issuance slows, whether share buybacks resume, and whether equity award levels compress for a third consecutive year.

If buybacks restart and equity grants stabilize, that is evidence the AI investment is generating enough return to ease the financing trade-offs. If debt issuance continues and equity cuts extend into a third year, the burden remains on both workers and the capital structure simultaneously, with the monetization case still unproven.

A company can report growing profits while cutting employee compensation, suspending shareholder returns, and borrowing at scale. Meta is currently doing all three. Each fact is accurate on its own. Together, they describe a company whose earnings headline and underlying cash dynamics are moving in opposite directions and whose workforce is absorbing compensation changes whose payoff Zuckerberg himself cannot yet map precisely.