Cloudflare's 2025 and 2026 acquisitions
What are the strategic implications of Cloudflare's 2025 and 2026 acquisitions (Replicate, Human Native, Astro, VoidZero)?
Taken together, the four deals are Cloudflare buying its way into the developer's workflow and into the business model for AI access. They add talent and influence over developers, not revenue. None is financially material except through stock-based pay.
What was bought
| Deal | Terms (source) | What it adds | Where it fits |
|---|---|---|---|
| Replicate (1 Dec 2025) | $57.4M cash. $46.6M goodwill, "primarily attributable to the assembled workforce". Acquired technology written off over 2 years (FY2025 10-K) | "An AI platform that enables developers to deploy and run AI models" (FY2025 10-K) | AI inference (Workers AI) |
| Human Native (announced by the Q4 FY2025 call) | Price not disclosed | A team from Google to design "the next business model of the internet … away from advertisements … away from subscriptions" (Prince, Q4 FY2025 call). Outside the sources, Human Native was a marketplace for licensing training data. | Act 4 (AI agents paying for content) |
| Astro (announced by the Q4 FY2025 call) | Price not disclosed | A web framework "powering a lot of Vibe-coding systems" (Prince, Q4 FY2025 call) | Developer platform (Workers) |
| VoidZero (1 Jun 2026) | $164.2M: $98.7M in stock (757,339 Class A shares), 106.4M of retention stock awards** vesting over about 3.9 years. $143.0M goodwill; $26.5M of customer relationships (Q2 FY2026 10-Q) | "High-performance toolchain technology" (Q2 FY2026 10-Q). Outside the sources, VoidZero is the company behind the Vite build tool. | Developer platform (Workers) |
Neither the FY2025 10-K nor the Q2 FY2026 10-Q prices Human Native or Astro separately. Acquisition cash beyond Replicate and VoidZero was about $6M in FY2025 and $10M in H1 2026, which suggests both deals were small (inferred).
Strategic implications
-
Cloudflare is moving up the developer stack, from where code runs to how code is written and built. Until now it competed at deployment, the last step. Owning a popular framework (Astro) and build toolchain (VoidZero) puts Cloudflare at the first step, where developers choose defaults. If the tools developers start with deploy most easily to Workers, the developer platform wins before a hyperscaler is considered (inferred). This is the developer side of the customer–ISP–developer loop, and it confirms that side is being built by purchase, not by network effect.
-
The target customer is AI-generated software, not only human developers. Prince tied Astro to "vibe-coding" platforms. Management has said most of those platforms already build on Workers or deploy to it by default (Q4 FY2025, Q2 FY2026 calls). If AI agents write more of the world's code, whoever owns the default templates and toolchains decides where that code is deployed (inferred). The acquisitions buy influence over those defaults.
-
Replicate fills a hole in the inference offer. Workers AI sells inference per request, but its value depends on which models customers can run. Replicate adds a platform for deploying and running a wide range of models. That partly hedges the new 10-Q risk that AI providers could stop their models running on Cloudflare (inferred). The 2-year life on the technology says Cloudflare bought people and time-to-market, not durable intellectual property.
-
Human Native is groundwork for Act 4, the only true two-sided market. Crawl controls and pay-per-crawl give Cloudflare the gate between publishers and AI companies. A team built around licensing content to AI supplies the marketplace logic on top. Nothing in the sources quantifies revenue, and management says it is "too early" to model (Q2 FY2025 call).
-
Acquisition pace and dilution are rising. VoidZero is Cloudflare's largest deal by headline price, above Area 1 ($156.6M in 2022). Including the retention awards, it costs about $270M, and roughly 75% of it is paid in shares (inferred from the 10-Q). That adds to stock-based compensation already running at about 21% of revenue. It also departs from the stated "very, very high" acquisition hurdle (Q1 FY2024 call), even though each deal on its own is small.
Risks these create
- Neutrality of open-source projects. Astro's and Vite's value comes from being usable anywhere. If Cloudflare visibly favours Workers, users could move to competing tools and the projects' influence would fade (inferred). This is the same neutrality argument Cloudflare uses against the hyperscalers, turned against itself.
- Keeping the teams. The value sits in people, with goodwill attributed to the assembled workforce and VoidZero's $106M retention vesting over about 3.9 years. The price is justified only if those people stay and stay productive.
- Low-margin mix. All four deals push toward developer products and inference, which already carry below-average gross margins (Q1 FY2026 call).
What to watch
- Whether the developer count (7.4M at Q2 FY2026) turns into disclosed Workers revenue. How many developers pay is not disclosed.
- Adoption of Astro and Vite outside Cloudflare, as a check that the projects stay neutral. There is no data in the sources.
- The first revenue from Act 4, which management named as a 2026 priority (Q1 FY2026 call).
- Stock-based compensation as a share of revenue once VoidZero's retention awards run through the income statement.
The Human Native and Vite descriptions are general knowledge, not taken from the sources.
