Lap splicing is the most common and cost-efficient method for connecting bars in GFRP-reinforced concrete structures. However, there is a lack of knowledge regarding the bond behavior of high-modulus GFRP bars in concrete. The present study aims to fill this gap by investigating the bond strength of lap-spliced high-modulus GFRP bars in concrete. These new-generation GFRP bars have a modulus of elasticity of up to 65 GPa and an ultimate tensile strength of up to 1400 MPa. Three large−scale splice beam tests, with a rectangular cross-section of 300 mm × 450 mm and a length of 5,200 mm, varying splice lengths (28d_b, 38d_b, and 45d_b), were designed to evaluate predictions of bond strength for North American design codes. Results show a nonlinear decrease in bond strength by 14.7% and 17.7% with a 36% and 60% increase in splice length, attributed to uneven bond stress distribution. On the other hand, these increases in splice length led to an increase in the splice strength by 17.7% and 31.8%, respectively. ACI 440.11-22 provides the most accurate predictions, showing an average test-to-prediction ratio of 1.03, while CSA S806-12 and CSA S6-19 overestimate with average test-to-prediction ratios of 0.84 and 0.57, respectively. In the latest Canadian Highway Bridge Design Code (CSA S6-25), the revised factor for GFRP-to-steel bond strength improves predictions with a test-to-prediction ratio of 0.89. However, Canadian codes generally over-predict bond strength, particularly by increasing the splice length to bar diameter ratio, neglecting the nonlinear bond stress distribution along the embedment length.