50%
Plausible
Apple is planning to skip the M6 Ultra launch, instead choosing to launch the M5 Ultra later this year, followed by the M7 Ultra after it unveils the M7 Pro and M7 Max sometime in 2027. Fortunately, the new workstation-class silicon will be worth the wait, especially for those who aspire to run multi-billion-parameter AI models on their systems but are limited by memory constraints. This is because a new report states that the M7 Ultra will feature double the unified memory of the M5 Ultra, and you’ll want to know how much RAM you can pack for the top-end configuration.
With the M5 Ultra reportedly limited to 768GB of unified memory, the M7 Ultra could overshadow it with a 1.5TB memory option
The ability to run advanced AI models depends on CPU speeds, with the heavy lifting being undertaken by the memory count and bandwidth. Fortunately, the M7 Ultra will address all three limitations as it’s expected to be the fastest Apple Silicon when it launches. While no CPU or GPU core counts are mentioned, Bloomberg’s Mark Gurman previously stated that the M5 Ultra will offer up to a 36-core CPU and an 80-core GPU, paired with 768GB of unified memory.
The M7 Ultra, on the other hand, will offer up to a mammoth 1.5TB of unified memory. Also, for those worried about running trillion-parameter models, we’ll remind you that the M3 Ultra featured 819GB/s of memory bandwidth, so we shouldn’t be surprised if the M7 Ultra can cross over to the 1TB/s bandwidth territory.
“The new Ultra is designed to support as much as 1.5 terabytes of memory — roughly double the capacity planned for the M5 Ultra — though whether Apple ultimately offers that configuration will depend on the state of the industry. Widespread memory-chip shortages have made the component harder to find and more expensive.”
The only obstacle that remains is how much the Mac Studio equipped with the M7 Ultra will cost. Given that the California-based titan is testing CXMT DRAM, the company should minimize its supply problem, but consumers should expect to pay around $20,000 for the maxed-out version.
Just compare the top-end configuration of the Mac Studio, which is kitted with an M3 Ultra coupled with 96GB of unified memory. The cheapest option retails for $5,299, assuming you’re willing to select fewer CPU and GPU cores, not to mention outfit the machine with just a 1TB SSD.
For those believing that the 1.5TB RAM on the M7 Ultra can’t handle one trillion-parameter models, the M3 Ultra with its 512GB RAM can run DeepSeek’s R1 AI model with 671 billion parameters. It should be possible for the M7 Ultra to run 1.2 trillion-parameter models with 8-bit quantization, so let us keep our fingers crossed for these future benchmarks.
News Source: Bloomberg
About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.
Follow Wccftech on Google to get more of our news coverage in your feeds.
