What percentage of a Proton's mass is potential/kinetic energy? So in an hydrogen atom, the total mass of the atom is equal to the masses of the proton, the electron, minus their net binding energy of around 13 eV. Making the total less massive than the sum of its parts by about 1 part in 100 million. As it also turns out, the electron follows the virial theorem, so that binding energy is actually in the form of -26 eV of electromagnetic potential energy and 13 eV of kinetic energy.
In a proton however, the total mass is equal to the masses of the three valence quarks plus the net binding energy, which is not only positive but accounts for 99% of the proton's mass. This is because the protons can never be in a free state, so while this binding energy is still positive, it is the minimum possible binding energy these quarks can have, and so attempting to dissassociate a quark increases the total mass of the system just like in the case for the hydrogen atom.
Now onto the question, what fraction of this total net binding energy can be considered as the potential energy of the gluon field between quarks and the gluons themselves, and what percentage can be considered to be in the kinetic energy of the quarks and gluons? The potential energy must still be negative (since it is still a potential well and the strong potential is attractive at  nuclear distances), but the virial theorem no longer holds (because the strong potential doesn't follow an inverse square law), so the kinetic energy can no longer be simply negative half of the potential (and that would lead to a negative net binding energy anyway, but we know from above its positive). The kinetic energy must therefore be some multiple greater than 1 of the potential energy, and I am wondering if one has calculated what that multiple is. No doubt some lattice QCD and supercomputer shenanigans are needed to get the result, but surely it has been done?
TL;DR, the ~929 MeV of the proton, ignoring the ~9 MeV of the valence quarks, is some amount of negative potential and positive kinetic energy, say -200 MeV potential and 1129 MeV kinetic, for example. Those numbers are what I'm looking for.
(Note, I am aware that in actuality most of this energy will be taking the form of virtual quark-antiquark pairs, but these are in constant flux and so I am merely looking for the semi-classical baseline that these quantum fluctuations float around.)
 A: There are a number of subtleties, because in a field theory you must define what you mean by "potential" and "kinetic" energy, and there are issues with any decomposition that are related to gauge invariance and scale dependence.
The issue of the proton mass decomposition has been studied in some detail, and continues to be an active area of research (see, for example, these two papers).
The basic idea is the following. Protons are described by QCD, and the QCD energy-momentum tensor is
$$
T^{\mu\nu} = \frac{1}{2} \bar\psi i\gamma^{(\mu}
\stackrel{\leftrightarrow}{D}\mbox{}^{\nu)}\psi
+\frac{1}{4}g^{\mu\nu}F^2 - F^{\mu\alpha}F_{\alpha}^{\nu}
$$
where $\psi$ is a quark field, $\stackrel{\leftrightarrow}{D}$ is a symmetrized gauge covariant derivative, $(\mu\nu)$ are symmetrized vector indices, and $F^{\mu\nu}$ is color field strength tensor. The Hamiltonian is
$$
 H = \int d^3x\, T^{00}
$$
and the proton mass is
$$
m_p = \langle p(\vec{k}=0) | H | p(\vec{k}=0)\rangle 
$$
where $| p(\vec{k}=0)\rangle$ is a proton state with zero momentum. I can now ask whether it is possible to decompose $m_p$ into contributions from various terms
in the stress tensor, and whether these terms can be independently measured.
The answer is "yes", modulo some of the ambiguities mentioned above (the total mass of the proton is certainly well defined, but there may be ambiguities in individual terms that cancel in $m_p$). Experimental input on individual terms comes from deep inelastic scattering (which measures trace free matrix elements of the quark and gluon energy-momentum tensor), and pion-nucleon scattering (which measures quark mass contributions). Individual matrix elements can also be determined in lattice QCD.
There are many details, discussed in the papers cited above. A typical result is shown in this figure (taken from a report on a future electron ion collider) which shows fractional contributions from quarks and gluons (with quark mass contributions separated out)

Postscript: Note that the proton is very different from a typical non-relativistic bound state. A non-relativistic bound state is made from some constituents with total rest mass energy $E_0=m_1c^2+\ldots + m_Nc^2$, and the statement that there is a bound state implies that $E=E_0-B$, where $B$ is biding is a (positive) binding energy. States of this type exist in the heavy quark sector of QCD (at least approximately), and are know as charmomium, bottomonium, etc.
However, the proton is not like that. It is made of approximately massless quarks and exactly massless gluons, but the proton mass is large, $m_p >> 2m_u+m_d$. There is no binding energy, the mass of the proton is positive energy of quark and gluon fields. Also, the proton cannot be ionized into quark constiutents. If I excite the proton, all I can do is generate additional quark-anti-quark pairs.
Second postscript: If you are really interested in an estimate, there are models of the nucleon (not quite QCD, but QCD inspired) in which one can talk about binding energy. For example, in the constituent quark model, quarks acquire an effective mass $m_Q\sim 400$ MeV from chiral symmetry breaking. Constituent quarks interact (by one gluon exchange, string potentials, or instanton induced fores) to make the proton, with a binding energy $3\times 400 - 935 \sim 265$  MeV.
Also, one can attempt to take the gluon field energy and split into an electric part ("kinetic energy") and a magnetic part ("potential energy"). This separation depends on the scale, but the magnetic part does indeed come out negative. A simple analysis (using numbers from the paper cited above) gives
$$
 \frac{1}{2}\langle p| E^2 p\rangle 
 \simeq 850\, {\rm MeV}\quad\quad
 \frac{1}{2}\langle p| B^2 p\rangle 
 \simeq -525\, {\rm MeV}
$$
A similar separation cannot be done for quarks, because there is no gauge invariant way to decompose the covariant derivative $\bar\psi \gamma\cdot D\psi$ into a kinetic and an interaction term.
A: I took this image of the  from the very instructive webpage of Matt Strassler. It shows impressive complexity. It illustrates that the sum of the rest masses of the naive quark composition uud, 9.4 MeV/c$^2$, is only 1% of the proton mass of 938 MeV/c$^2$. The rest of the proton mass, according to Strassler, is quark and gluon kinetic energy and interaction energy. In this sense 99% of the proton mass is kinetic and potential energy.
"if the up and down quark masses were zero (and everything else was left unchanged), the proton mass would barely change from what we observe it to be."
A: 9 % of proton mass is potential and 91 % is kinetic energy, see for more details article.
Original publication
