A 2.5-3σ discrepancy has been reported between the baryonic acoustic oscillation peak (BAO) in the Lyman α forest at z ∼ 2.34 and the best fit Planck ΛCDM cosmology. To isolate the origin of the tension, we consider unanchored BAO, in which the standard BAO ruler is not calibrated, eliminating any dependence on cosmology before redshift z ∼ 2.34. We consider BOSS BAO measurements at z ∼ 0.32, 0.57 and 2.34, using the full 2-dimensional constraints on the angular and line of sight BAO scale, as well as isotropic BAO measurements by 6dF and SDSS at z ∼ 0.106 and z ∼ 0.15. We find that the z > 0.43 data alone is in 2.9σ of tension with ΛCDM with or without the Planck best fit values of the mass fraction Ω m and the BAO scale r d H 0 , indicating that the tension arises not from the ΛCDM parameters but from the dark energy evolution itself at 0.57 < z < 2.34. This conclusion is supported when the acoustic scale measured by the CMB is included, which further increases the tension and excludes a solution with a constant dark energy equation of state. Including the low z BAO data, which is itself consistent with ΛCDM, reduces the tension to just over 2σ, however in this case a CPL parametrization of the dark energy evolution yields only a modest improvement.