Haynes King broke a 20-20 tie in Boulder with a 45-yard touchdown run with 1:07 remaining, and Georgia Tech escaped the programs’ first meeting with a 27-20 win. That play settled the 2025 opener; it also captures exactly what Tech no longer has available in Thursday night’s rematch.
King, the 2025 ACC Player of the Year, is gone. After spring practice, Brent Key identified Indiana transfer Alberto Mendoza as the likely starter if Tech had played then, while leaving room for the competition to change during fall camp. The distinction matters. Mendoza is the expected answer, not an established one.
Georgia Tech vs. Colorado
Thursday, Sept. 3, 8 p.m. ET, ESPN
Rooting interest. Georgia Tech
Emotional investment. High in the performance, measured in the result
The Fan Watch model gives Georgia Tech a 78% chance to win at home. That is substantial favorite territory. A loss would be a real opening-night failure, but merely surviving would leave the important question unanswered.
Watch whether Mendoza can operate the whole offense without Godsey reducing it to a shelter. Mendoza’s reserve work offers legitimate encouragement. He completed 18 of 24 passes for 286 yards, five touchdowns and one interception for Indiana in 2025. But those numbers came without a college start. Thursday is about how that efficiency holds up when every meaningful decision belongs to him.
The best bridge is the backfield. Malachi Hosley returns as an All-ACC running back, and Michigan transfer Justice Haynes produced 857 rushing yards, 10 touchdowns and 7.1 yards per carry across seven games last season. An injury kept Haynes out of Michigan’s final four games, so his pregame availability deserves confirmation rather than assumption.
If that group lets Mendoza play decisively instead of cautiously, Tech can preserve the breadth of an offense that averaged 461.2 yards per game in 2025. The useful signs are not just handoffs and possession. Look for Mendoza to keep the ball when the read demands it, attack beyond the safest throws and prevent Colorado from crowding the backfield. Godsey does not need to reproduce King in one night; he does need to show that losing King has not made the offense small.
There is a physical benchmark, too. Georgia Tech survived turnovers on its first three possessions in Boulder and still finished with a 320-146 rushing advantage. Repeating that control would make Mendoza’s transition far less precarious. Failing to approach it would place much more of the game on the first-time starter than a 78% favorite should require.
The model has no outside result carrying meaningful Georgia Tech value, which leaves one unusually precise tension for a heavy favorite. The Jackets should control Colorado without asking Mendoza to imitate King. Root for the backfield to keep the full playbook available, then judge the important snaps by whether Mendoza is trusted to resolve them rather than protected from them.
The full set of probabilities is in the interactive Georgia Tech guide, with the simulation approach explained on the methodology page. The satisfying outcome is not a reenactment of King’s escape in Boulder. It is a game that never needs one.