Apple pushed out the base-tier M6 and high-end M5 Ultra chips on August 25, 2026, launching new Mac mini and Mac Studio configurations that go on sale September 22, 2026, according to Apple’s newsroom announcement and coverage from 36Kr. However, industry reporting indicates Apple is skipping the M6 Pro, Max, and Ultra tier to focus on an AI-optimized M7 family and a server-only chip called Baltra developed with Broadcom.
What Actually Shipped in Apple’s August 2026 Release
Apple’s newsroom announcement highlights the M5 Ultra as its primary hardware debut. The chip carries a next-generation GPU with up to 80 cores, each equipped with a Neural Accelerator. According to Apple, this delivers up to 4.5 times the peak GPU AI compute of the M3 Ultra and over 6 times that of the original M1 Ultra. The GPU also features third-generation ray tracing and runs up to 40% faster in graphics workloads than the M3 Ultra, based on Apple’s figures.
The base-tier M6 chip introduces a different architectural approach. 36Kr reports that M6 is the first Apple silicon chip to feature two separate 16-core Neural Engine blocks called simultaneously by the system framework, doubling peak AI compute over the previous generation. 36Kr also notes that the refreshed Mac Studio ships in M5 Max and M5 Ultra configurations, priced from 19,999 yuan and 46,999 yuan respectively in China, with all four new desktop configurations (M6, M5 Pro, M5 Max, M5 Ultra) hitting shelves on September 22, 2026.
Did you know? While Apple’s desktop refreshes arrive on September 22, 2026, reporting from outlets like TechSpot indicates that the newly announced M5 Ultra chips are also being deployed temporarily in Apple’s own server racks to bridge the gap before dedicated server silicon arrives.
Why Apple Is Skipping the M6 Pro, Max, and Ultra Tier
Apple is reportedly breaking its traditional hardware release cadence established since the 2020 M1 launch. Reporting from Bloomberg’s Mark Gurman, relayed by Gadgets Now and MacDailyNews, indicates Apple will ship only a base M6 chip this year for entry-level Macs and skip M6 Pro, M6 Max, and M6 Ultra entirely.
Instead, Apple is fast-tracking a full M7 lineup built specifically for artificial intelligence workloads. Gizchina’s roadmap analysis describes M6 as a transitional stopgap while Apple redirects engineering resources toward the AI-centric M7 architecture.
Inside the AI-First M7 Family and Upgraded Specs
iPhone in Canada coverage details a phased deployment schedule for the M7 collection: the base M7 variant arriving in the initial half of 2027, the M7 Pro and M7 Max editions concluding 2027, and the M7 Ultra following in 2028.
CryptoBriefing notes that the base M7 targets around 240 GB/s of memory bandwidth. Meanwhile, Tech Times reports that the 2028 M7 Ultra is planned to support up to 1.5 TB of memory—double the capacity of current M5 Ultra machines—allowing users to run massive local AI models without relying on dedicated GPU servers.
Baltra: The Server-Only AI Chip Built With Broadcom
Running parallel to the M-series is a server-only project codenamed Baltra. Analyst Ming-Chi Kuo’s insights featured in TechSpot indicate that Apple targets mass production for Baltra during late 2026, with the chips entering service across Apple-operated data centers throughout 2027.
Baltra is designed strictly for Apple’s own infrastructure to handle Apple Intelligence workloads rather than shipping in consumer Macs. Bloomberg-based reporting indicates these server chips could feature between double and eight times the CPU and GPU core counts of the current M3 Ultra.
Chasing Nvidia’s Blackwell in the 2026 AI Hardware War
Tech Times frames the M7 Ultra server plan as a direct challenge to Nvidia’s Blackwell-generation GPUs. While Nvidia dominates data center throughput via CUDA software lock-in, Apple is competing on unified memory capacity and power efficiency per query.
This strategy mirrors an industry-wide processor war. A July 2026 analysis published by Nate News highlights that Intel, AMD, and Qualcomm are fighting for mainstream Windows AI PC dominance, while Nvidia focuses on high-end systems and Apple pursues a vertically integrated consumer and data center lane.
Pro Tip for Mac Buyers
If you need a machine immediately, the M6 and M5 Ultra offer genuine performance upgrades, with M5 Ultra delivering up to 4.5x higher peak GPU AI compute than the M3 Ultra. Because the M7 family is not expected until 2027 and M7 Ultra until 2028, waiting means delaying your purchase by 18 to 24 months.
Frequently Asked Questions
What did Apple actually announce on August 25, 2026?
Apple announced the base-tier M6 and high-end M5 Ultra chips, powering refreshed Mac mini and Mac Studio configurations that go on sale September 22, 2026, according to Apple’s newsroom and 36Kr.
Is Apple skipping M6 Pro, M6 Max, and M6 Ultra?
Yes, multiple reports including Bloomberg-sourced coverage via Gadgets Now and MacDailyNews state that Apple is skipping higher-end M6 variants to focus directly on an AI-focused M7 family.
What is Baltra?
Baltra is a reported Broadcom-developed AI server chip project intended exclusively for Apple’s own data centers, with mass production slated for the second half of 2026 and data centers launching in 2027, according to TechSpot and analyst Ming-Chi Kuo.
When is the M7 chip family expected?
Reporting cited by iPhone in Canada places the base M7 in the first half of 2027, M7 Pro and Max by late 2027, and M7 Ultra in 2028.
How much memory will M7 Ultra support?
Tech Times reports that M7 Ultra is targeting up to 1.5 TB of memory, doubling current M5 Ultra ceilings to run large local AI models.
Should I buy an M6 or M5 Ultra Mac now, or wait for M7?
Buyers needing hardware now can benefit from shipping M6 and M5 Ultra performance gains, as M7 configurations will not launch until 2027 and 2028.
How does Apple’s AI chip strategy compare to competitors?
While Intel, AMD, and Qualcomm target Windows AI PC market share, and Nvidia focuses on data center GPUs and high-end systems, Apple utilizes vertical integration combining consumer Macs with in-house server silicon.
Will Apple sell AI server capacity to outside developers?
Current reporting contains no indication that Apple plans to offer Baltra or M7 Ultra server capacity as a public cloud product, keeping the infrastructure dedicated to Apple Intelligence.
Explore more tech insights and industry news by subscribing to our weekly newsletter, and join the conversation in the comments below.
Related reading