At its Intel Foundry Direct Connect 2024 event, the American company confirmed its ambition to compete head-on with TSMC’s foundries. First of all, the sprint launched by Pat Gelsinser to bring his production tool up to the Taiwanese’s level is set to gather pace. When the blue-chip boss announced his strategy of launching 5 manufacturing processes in 4 years, it seemed impossible… However, in reality, the gamble seems on the point of being won (Intel 20A and Intel 18A nodes should go into production during 2024). But there’s more to come. By 2027, Intel now wants to launch 7 new nodes. A crazy schedule for the next 3 years! If this plan comes to fruition, Intel will once again be the world’s second-largest foundry by 2030.
Gelsinger keeps up the pressure on its chip foundries
By mixing business, politics and AI, Intel also wanted to make its mark during its conference. Like a fellow American tech company, Intel also threw itself into “trendy” marketing concepts, selling its audience on a more sustainable systems foundry business designed for the AI era . All this before a very receptive audience that included ARM CEO Rene Haasn, US Commerce Secretary Gina Raimondo, and the man of the moment, Open AI CEO Sam Altma.
Node 18A: the moment of truth approaches for Intel
But the economic and technical moment of truth should come in 2025. At that time, as mentioned above, the Intel 18A process will be launched, and should enable Intel to regain its technological leadership in chip manufacturing. This leadership is not merely symbolic, given the accelerating demand for complex chips, particularly those required by AI.
Partnership with UMC and Tower Semi for volume production
But Intel doesn’t stop at the most advanced processes. During the conference, close partnerships were announced with UMC and Tower Semiconductor (which the blueprints once wanted to buy out). The idea is to collaborate with these two companies to bring to market chips with mature processes (65 nm in particular) aimed at companies who want simple, low-cost chips in very high production volumes.
