A homogeneous dust sphere begins to contract under its own gravity. Inside it, like a rescue bubble, a region of negative pressure inflates: the de Sitter bubble. It expands, slows down near the very edge of the fall, and freezes at the Schwarzschild radius. This is how a gravastar is born: a compact object without a horizon or singularity.

The Oppenheimer–Snyder collapse, the contraction of homogeneous dust into a black hole, serves as the starting point. Into the collapsing dust, an expanding de Sitter bubble with negative pressure is introduced, initially of zero size. Spacetime is divided into three regions: the inner de Sitter region, the intermediate dust region, and the outer Schwarzschild region. Matching conditions at the boundaries ensure the continuity of the metric, and on the de Sitter boundary, surface tension and surface energy density arise. Instead of directly simulating from initial conditions, the authors integrate the equation backward in time from the final gravastar state, which allows them to find the necessary fine-tuning of parameters.

The parameter space is divided by a separatrix: a boundary between different outcomes of evolution. Below it, collapse leads to a black hole; above it, to a non-equilibrium configuration; and on the line itself, to a gravastar. Each point on the separatrix corresponds to an infinitely fine tuning, but there are infinitely many such points. This gives an infinite family of initial conditions for gravastar birth. A compactness limit is found: for successful gravastar formation, the initial compactness of the dust sphere must satisfy C = M/R̄₂ ≤ 3/8 = 0.375, which is only slightly less than the Buchdahl limit of 4/9 ≈ 0.444. For greater compactness, collapse inevitably ends in a black hole.

The work demonstrates for the first time that within classical general relativity, without higher curvatures or quantum corrections, a collapse is possible that ends neither in a singularity nor in an event horizon. This differs from neighboring works, where the formation of horizonless objects was either postulated or required modified gravity. Here, the minimal extension of the Oppenheimer–Snyder collapse is used by adding a de Sitter bubble, making the model transparent and testable. The result raises the question of how realistic the fine-tuning required for gravastar birth is, and whether it can arise naturally, for example, from quantum fluctuations or a conformal anomaly.

A rescue bubble inflating inside a collapsing star and freezing at the very edge of the fall: that is how one might describe the birth of a gravastar. In the future, it is necessary to check the stability of the obtained objects against non-radial perturbations and to investigate more realistic equations of state. Perhaps the fine-tuning will turn out to be less stringent when quantum effects are taken into account. The question remains open as to which initial conditions in a real collapse are more likely: those leading to a black hole or to a gravastar.