Major Hardware Update: Apple Upgrades Mac mini and Mac Studio

Contents

Unexpected hardware bombshell from Cupertino: Apple is surprisingly announcing a major update to the Mac mini and Mac Studio, along with two new processors. New are the M6 and M5 Pro in the Mac mini, and the M5 Max and M5 Ultra in the Mac Studio. Apple claims significant performance gains compared to the previous models with M4, M4 Pro, M4 Max, and M3 Ultra. Particularly in focus: local AI. The target audience is likely to be enterprise customers.

The timing of the announcement is unusual: In previous years, Apple reserved new Macs for the fall. The fact that new models are being introduced in August instead of October or November is due to high demand, according to sources in Cupertino. The use of AI agents like OpenClaw, in particular, has driven strong demand for Apple. This was accompanied by a need for powerful successors. Pre-orders for the new Macs are possible starting today, with sales scheduled to begin on September 22.

M6 ushers in the 2nm era

It is remarkable how Apple’s version designations are diverging with the new Macs. While the M3 Ultra was previously the outlier of the M4 line, the M6 is now sprinting past the M5. The M6 is the first Apple chip manufactured using a 2-nanometer process. The M5 Pro, M5 Max, and M5 Ultra remain at the 3-nanometer process node. The M5 Ultra, which connects two M5 Max dies via UltraFusion, has a significantly improved interconnect bandwidth of 4.4 TB/s – for comparison, the M3 Ultra had 2.5 TB/s.

The Mac mini with M6 makes a two-generation leap in chips. The CPU has 12 cores (2 Super-Cores, 4 Performance-Cores, 6 Efficiency-Cores) and is said to be up to 40 percent faster in multi-threading than the M4. The 12-core GPU brings Neural Accelerators to the Mac mini for the first time and is expected to provide up to 4x higher AI computing power and twice as fast graphics performance compared to the M4. Apple has also significantly increased the Neural Engine. It is a dual 16-core Neural Engine. The Unified Memory with up to 170 GB/s bandwidth also shows an increase compared to the M4. The M6 can be equipped with up to 32 GB of Unified Memory.

Mac mini with M5 Pro is cluster-capable

For the Mac mini with M5 Pro, Apple promises up to four times faster AI workloads. The chip debuted in March 2026 in the MacBook Pro. It can be ordered in the Mac mini with up to 64 GB of Unified Memory. Further improvements – for both the Mac mini with M6 and the M5 Pro – include up to twice as fast storage, the integration of Apple’s own N1 chip for Wi-Fi 7 and Bluetooth 6, and a new standard 2.5 Gigabit Ethernet port (previously 1 Gbit/s). The LAN port can also be upgraded to 10 Gbit/s for an additional charge. The cooling system remains the same. Base configurations still start at 16 GB.

An important difference: The Mac mini with M6 has three Thunderbolt 4 ports, while only the M5 Pro gets Thunderbolt 5 – and thus the cluster capability for local AI models.

Ultra chip with the biggest leap since the beginning

The Mac Studio with the M5 Max, introduced in the spring in the MacBook Pro, also makes a significant performance leap over its predecessor M4 Max: The 18-core CPU is said to be up to 30 percent faster, the 40-core GPU up to 50 percent. Neural Accelerators are now integrated into every single GPU core, which Apple claims enables up to four times faster AI tasks. It can be configured with up to 128 GB of Unified Memory and has a memory bandwidth of up to 614 GB/s.

The bigger leap is marked by the M5 Ultra, the successor to the M3 Ultra. The 36-core CPU (12 Super-Cores, 24 Performance-Cores) is expected to deliver up to 30 percent more multi-thread performance compared to the M3 Ultra. The 80-core GPU brings Neural Accelerators to the Ultra chip for the first time, achieving up to 4.5 times higher AI computing power, along with a redesigned shader core and a third-generation ray tracing engine for up to 80 percent more graphics performance. Apple has also significantly increased Unified Memory: bandwidth increases by 50 percent to 1.2 TB/s, with a maximum capacity of 512 GB. For workflows with particularly high computing demands – such as training AI models or running models with over a trillion parameters – multiple Mac Studios can be clustered via Thunderbolt 5. Apple uses RDMA (Remote Direct Memory Access) for this purpose.

The UltraFusion bandwidth leap in the M5 Ultra is the first real performance increase in the interconnect since the technology was introduced in 2022. This is likely why Apple skipped the M4 Ultra. The M5 Ultra has also seen significant improvements in the Media Engine. It can handle up to 33 simultaneous 8K ProRes 422 streams at 30 fps.

Prices and Availability

Two hurdles that the major upgrade of the devices does not lower are prices and availability. Apple had significantly increased its prices a few weeks ago due to memory shortages. The Mac mini M6 in its base configuration will cost 1049 Euros – significantly above the starting price of the Mac mini M4 two years ago, which was 699 Euros. And the price has now been increased again, as the Mac mini with M4 recently cost 949 Euros. The Mac mini M5 Pro is available from 1999 Euros, while the Mac mini M4 Pro was still 1649 Euros at launch. The Mac Studio with M5 Max is available from 2999 Euros (Mac Studio M4 Max at market launch 2499 Euros). The price difference for the Mac Studio M5 Ultra compared to the market launch of the M3 Ultra is very significant. Instead of 4999 Euros, prices start at 6599 Euros.

Apple says it will produce as many devices as it can – meaning the expected demand is likely to exceed supply, potentially leading to waiting times. The variant of the Mac Studio equipped with 512 GB of Unified Memory will generally only be available from the end of October.

Focus on Enterprise Customers

While the prices might make ambitious private customers hesitate, it might look different from the perspective of some enterprise customers: They can offset the massive performance increase of Apple’s “AI powerhouses” against the tokens they save by opting for on-device AI instead of cloud providers. Further arguments for local AI are data protection and data sovereignty. Apple, apparently surprised by the strong AI-driven demand for its hardware itself, now apparently wants to seize the moment and push the AI agenda forward with significant performance gains. The switch to 2nm for the M6 is also a sign. The biggest challenge for Cupertino currently is likely not the prices, but whether it can meet the demand.

(mki)