Manufacturing cost plays a crucial role in suppliers’ encroachment decisions. A high manufacturing cost impedes suppliers’ capacity to encroach. However, cost learning may reduce this cost sufficiently enough to make encroachment profitable for the supplier at a later point in time. Accordingly, he may have an incentive to boost production so as to promote cost learning. Thus, he may drop the wholesale price to induce the retailer to buy more. On the one hand, cost learning may enable encroachment, which may be detrimental to the retailer. On the other hand, cost learning results in a lower manufacturing cost which may translate into a lower future wholesale price, benefiting the retailer. Therefore, the retailer faces a dilemma: should she increase her order quantity to advance cost learning or not? As the retailer may order fewer units in the initial period to limit future direct channel sales, the supplier faces a challenge: should he, instead of dropping his initial wholesale price, raise it to signal his intention of not encroaching so as to induce the retailer to sell a higher quantity in the first period? We model the supplier–retailer interaction as a two-period Stackelberg game to address the retailer’s dilemma and to identify the optimal supplier response. We uncover a new outcome, which arises in the presence of cost learning, where the supplier encroaches but decides not to sell anything through the direct channel. In addition, we find that supplier encroachment may reduce or eliminate the retailer’s incentive to advance cost learning. This results in lower sales by the retailer, which impedes cost learning, leading to a higher future manufacturing cost (compared to the no encroachment setting). As a result, encroachment, which is typically viewed as advantageous for the supplier, may become detrimental to him. Surprisingly, the supplier continues to encroach and sell directly unless he can credibly assure the retailer that he will not encroach in the future.