Here is the first image purportedly representing an impending NVIDIA flagship or “halo” device that will sit above the GeForce RTX 4090. The GeForce RTX 4090 Ti and the NVIDIA RTX TITAN are two different possible brand names for this item (Ada). NVIDIA has ample of leeway to create a halo product that fully utilizes the 144 streaming multiprocessors (SM) on the 4 nm “AD102” silicon because the RTX 4090 only employs 128 of them (88 percent). In addition to pushing the silicon to its limits, NVIDIA has the chance to use faster 24 Gbps-rated GDDR6X memory chips and raise the usual graphics power closer to the 16-pin ATX 12VHPWR connector’s 600 W continuous power delivery maximum (the RTX 4090 uses 21 Gbps memory).

With the rear I/O bracket spanning all 4 slots, the card is 4 slots thick. Instead of being placed along the base, the display outputs of the card are arranged along its thickness. The Dual-Axial Flow Through cooler from the RTX 4090 Founders Edition was massively scaled up to create the cooler. As opposed to other add-on cards, which have their PCBs perpendicular to the motherboard plane, this card’s PCB runs parallel to the motherboard plane. The PCB is organized along the card’s thickness. This was probably done to maximize the amount of space that the cooling system took up and perhaps even to accommodate a third fan. We also anticipate that the PCB will be divided such that a smaller PCB will handle the display I/O while a third PCB will take care of the PCI-Express slot interface. It is sufficient to say that NVIDIA’s RTX 4090 Ti / RTX Titan will be a work of engineering art.