Something shifted in the composition of technology sector returns that deserves careful attention from advisors and long-term investors. In a September 2026 report published on Platform Aeronaut1, Thomas Reiner, a researcher affiliated with Altimeter Capital, documents what four consecutive quarters of rising run-rate dilution actually look like across software and internet companies and, crucially, why a superficially clean dilution headline can obscure a deeply problematic capital structure.
The argument is straightforward but analytically layered. Reiner had launched his online dilution tracker in March 2026, at a point when the data seemed encouraging. As he writes, "we had seen 1.5 years of declining dilution and although there were some outliers, generally companies were taking their medicine." That period of rationalization, it turns out, was temporary. Dilution peaked in the fourth quarter of 2022 through early 2023, driven largely by grants sized at inflated COVID-era valuations. As those grants rolled off, companies reduced headcount and stock prices recovered. The trend looked disciplined. Then came the AI supercycle. "With the surge of AI and declining stock prices," Reiner notes, "that rationalization is dead."
The Stock Price Blind Spot
The central analytical contribution of Reiner's report is the per-employee lens. Traditional dilution metrics track share counts, but they carry a hidden distortion: companies and employees alike think about equity compensation in dollar terms per person, not in shares. If the stock doubles, the same grant plan looks disciplined. If the stock drops by half, it looks reckless. The share count moves, but the intent did not. To strip out that noise, Reiner has added headcount and per-employee metrics to every company in his tracker, enabling a comparison of equity grants per head against revenue per head and free cash flow per head. The resulting picture is more revealing than any headline dilution percentage.
Across the coverage universe, year-over-year increases in grants per employee are significant. Software companies are granting $88,000 per employee versus $68,000 a year ago, a 29% increase. Internet companies have moved from $63,000 to $75,000, up 19%. Big Tech shows the most dramatic shift: from $41,000 to $69,000 per employee, a 68% increase in a single year. Reiner attributes much of this to the competitive pressure of the AI talent market, with OpenAI and Anthropic drawing away high-quality engineers from the broader technology sector.
Winners and Losers Are Separating
The per-employee scatter plot Reiner presents is instructive. Mega-cap companies generate approximately $1 million in revenue per employee and grant roughly $69,000 of stock per head. Smaller-cap companies generate only $547,000 in revenue per employee but grant $63,000 in equity, nearly the same absolute dollar amount. As Reiner observes, "both sides of the equation move the wrong way at once as you go down the market cap scale: revenue per head falls by nearly half, and equity per head doubles." That arithmetic explains why small-cap dilution runs at 4.1% against 1.1% for mega caps, and why, as he dryly notes, "we'll grow into it does so much work in small-cap investor decks."
Reiner offers Nvidia as the counter-example that clarifies the principle. The company grants $186,000 of stock per employee, six times its mega-cap cohort. The number is large. But Nvidia generates $7.8 million in revenue and $3.3 million in free cash flow per employee. The grants are, as Reiner writes, "enormous in absolute terms but trivial relative to what each employee produces." Generous compensation is not the problem. Generous compensation the company and its workforce have not earned is the problem. Reiner frames the conclusion plainly: "We're increasingly moving into a world of winners and losers as we get deeper into the AI supercycle."
Cloudflare presents a more ambiguous case. Its blended dilution of 0.8% sits below the large-cap cohort on every headline measure, and looks disciplined at a glance. But the per-employee data tells a different story: SBC per employee runs at $109,000, roughly 6.5 times cohort, against $67,000 of FCF per employee. The company is paying each employee approximately $1.60 in stock for every $1.00 of free cash flow that employee generates. Reiner puts a sharp point on it: "Nothing in the dilution rows is wrong, the stock price is simply high enough that a very large equity budget dissolves into a small share count."
At the other extreme, Snapchat is running at more than 10% annual dilution and granting 1.7 times more equity to employees than the company generates in free cash flow. Headcount reductions are underway, but as Reiner states, they are simply "not enough as the health of the business doesn't support the compensation provided to employees (or management)." C3.ai is generating negative free cash flow and granting nearly as much in equity as it earns in revenue, a clearly unsustainable configuration. Pinterest's forward dilution stands at 8.6% and is accelerating, with grants per employee at more than four times its mid-cap cohort.
A Three-Question Checklist
Reiner closes with a practical framework. Per-employee data adds three questions to any dilution analysis. First, is the headline dilution number being flattered by the stock price? If the dilution rows appear clean but SBC per employee is running at a multiple of the cohort, the stock is doing the work, not compensation discipline. Second, can the business actually afford its grants? SBC per employee running below 30 to 40% of FCF per employee suggests the grants are funded by operations. Approaching 100%, the company is returning every dollar of cash generation to employees as equity. Above 100%, or with negative FCF per head, the company is borrowing from shareholders to pay its workforce. Third, where is dilution heading? Reiner's "dilution acceleration" metric, calculated as forward minus trailing dilution, gives the directional signal. Pinterest and Wayfair sit at plus 5.6 and plus 4.8 respectively. Carvana, a notable positive example in this report, sits at negative 3.3, meaning its worst dilution is behind it.
Five Key Takeaways for Advisors and Investors
- Four consecutive quarters of rising dilution signals a regime shift. The 2022-2023 rationalization in tech compensation was real but temporary. AI-driven talent competition has reversed that trend broadly across software and internet sectors, and the reversal is measurable and ongoing.
- Headline dilution percentages can be misleading. A low dilution percentage may simply reflect a high stock price absorbing a large equity budget. Per-employee analysis strips out that distortion and reveals whether compensation discipline is genuine or stock-price-dependent.
- Small-cap tech carries a structurally higher dilution burden. Revenue per head falls sharply as market cap declines, while equity per head barely moves. The resulting 4.1% average small-cap dilution rate versus 1.1% for mega caps is not incidental; it is built into the economics of smaller-scale operations competing for the same talent pool.
- Free cash flow per employee is the relevant affordability benchmark. Whether a company can sustain its equity grants is answered by comparing SBC per head to FCF per head. Companies running above 100%, or generating negative FCF per head, are transferring wealth from shareholders to employees in a manner the business has not earned.
- AI is accelerating the divergence between strong and weak capital allocators. Nvidia and Netflix demonstrate that generous equity compensation is a positive signal when FCF per employee is high. Snapchat, C3.ai, and Pinterest demonstrate the inverse. The AI supercycle is not lifting all boats; it is sharpening the distinction between companies whose compensation structures are sustainable and those whose are not.
Footnote:
1 Reiner, Thomas. "Tech Dilution is Creeping Up Again + The Per-Employee Lens." Platform Aeronaut, 22 Sept. 2026, https://www.platformaeronaut.com/p/tech-dilution-is-creeping-up-again.
