JWST is probably less likely to personally “kill” Λ than to make the error bars stop pretending they’re decorative. Its big contribution is getting better distances, galaxy ages, star-formation histories, and gravitational-lensing maps at high redshift. That helps test the model, but dark energy is mostly constrained through the expansion history, so JWST isn’t going to stare at a galaxy and catch Λ changing its alibi.
The more dangerous scenario is a disagreement that appears in several independent probes at once: supernova distances, BAO, weak lensing, cluster counts, and the growth of structure all pointing to the same evolving equation of state. One odd early galaxy can be blamed on dust, calibration, selection effects, or astronomers once again underestimating how messy galaxy formation is. A coordinated mismatch across unrelated surveys is much harder to hide under the rug.
And as Fergie said, Λ has the unfair advantage of being absurdly economical. It’s basically one constant that fits an embarrassing amount of data, which is the scientific equivalent of a cheap appliance that somehow survives every move. DESI, Euclid, Rubin, Roman, and future CMB experiments will probably matter more for cornering dark energy than JWST alone. The real question is whether any replacement explains the anomalies without requiring five new parameters and a ceremonial sacrifice to Occam’s razor.