Apple has introduced two new processors, the M5 Ultra and M6, as it expands its custom silicon strategy around high-performance computing and on-device artificial intelligence. The M5 Ultra, which Apple calls its most powerful chip ever, will power the new Mac Studio, while the M6 debuts in the latest Mac mini. Both chips are designed to deliver significant gains in computing and AI workloads.
The launches come as Apple places greater emphasis on running AI workloads locally rather than relying entirely on cloud-based processing. The M6 is Apple’s first chip built using a 2-nanometer process, while the M5 Ultra introduces Apple’s first quad-die architecture in an M-series system. The company is positioning the two processors for different segments, with M6 aimed at everyday and developer workloads and M5 Ultra targeted at demanding professional applications and large AI models.
M5 Ultra Becomes Apple’s Most Powerful Chip
The M5 Ultra is the flagship of Apple’s latest silicon lineup and is designed for the new Mac Studio. Apple has created the chip by combining two dual-die M5 Max chips using its UltraFusion interconnect technology, producing the first four-die architecture in an M-series chip.
The architecture allows the four dies to operate as a single processor. Apple says UltraFusion provides more than 4.4 TB/s of inter-die bandwidth and more than six times the connection density of its previous approach.
M5 Ultra Key Specifications
| Feature | M5 Ultra |
|---|---|
| CPU | Up to 36 cores |
| GPU | Up to 80 cores |
| Neural Engine | 32 cores |
| Unified memory | Up to 512GB |
| Memory bandwidth | Up to 1.2TB/s |
| Architecture | Quad-die |
| Inter-die bandwidth | More than 4.4TB/s |
| Primary Mac | Mac Studio |
The M5 Ultra has up to 36 CPU cores, including 12 super cores and 24 performance cores, and up to 80 GPU cores. It supports as much as 512GB of unified memory and delivers up to 1.2TB/s of memory bandwidth.
Major AI Performance Improvements
AI is central to Apple’s positioning of both new processors. The M5 Ultra’s GPU includes Neural Accelerators, allowing the chip to handle AI-focused workloads more efficiently.
Apple says peak GPU compute for AI applications is up to 4.5 times higher than the M3 Ultra. The company is targeting workloads including large AI models, 3D rendering, visual effects and professional video production.
The chip’s large unified memory capacity is particularly important for developers working with large language models. Instead of relying exclusively on external cloud infrastructure, users can run increasingly capable AI models locally on a Mac Studio.
M5 Ultra Performance Compared With M3 Ultra
| Metric | M5 Ultra vs. M3 Ultra |
|---|---|
| Single-threaded CPU performance | Up to 1.25x |
| Multithreaded CPU performance | Up to 1.3x |
| AI peak GPU compute | Up to 4.5x |
| Unified memory bandwidth | 1.2TB/s |
| Memory capacity | Up to 512GB |
| Inter-die bandwidth | More than 4.4TB/s |
The 1.2TB/s memory bandwidth is also 50% higher than the M3 Ultra, according to Apple.
M6 Is Apple’s First 2nm Mac Chip
While the M5 Ultra is focused on maximum performance, the M6 represents Apple’s next step in chip manufacturing and mainstream Mac performance.
The M6 is Apple’s first 2-nanometer processor for its Mac lineup. Apple says the smaller process allows it to increase performance and efficiency while adding more computing resources.
The chip features a 12-core CPU, a 12-core GPU and a Dual 16-core Neural Engine. Its CPU adds two cores compared with the M5 and uses a configuration consisting of two super cores, four performance cores and six efficiency cores.
M6 Key Specifications
| Feature | M6 |
|---|---|
| Manufacturing process | 2nm |
| CPU | 12 cores |
| GPU | 12 cores |
| Neural Engine | Dual 16-core |
| Unified memory | Up to 32GB |
| Memory bandwidth | Up to 170GB/s |
| Primary Mac | Mac mini |
Apple says M6 provides the world’s fastest single-threaded performance and up to 1.2 times faster multithreaded performance than M5. It also delivers nearly 30% higher peak GPU compute for AI compared with M5, according to the company.
M6 Targets Local AI Developers
The M6’s AI improvements are significant because Apple is increasingly presenting the Mac as a platform for developing and running AI applications locally.
The chip’s Neural Engine and GPU Neural Accelerators are designed to accelerate AI workloads without requiring every request to be sent to a cloud-based server. Apple says developers can use its frameworks to run and fine-tune large AI models locally.
Apple’s software ecosystem also allows developers to combine its Foundation Models, App Intents and proprietary AI models to create applications that operate on the device.
This could become increasingly important as AI agents require faster responses and greater access to local files and applications. Running at least some AI processing locally can also reduce dependence on network connectivity and potentially improve privacy.
New Mac Mini And Mac Studio Lead The Hardware Push
The two processors are debuting in different Mac products.
The M6 is available in the new Mac mini, which starts at $899. The Mac mini can also be configured with the M5 Pro for users who need higher performance. The new Mac Studio is available with M5 Max or M5 Ultra processors, with the M5 Ultra configuration starting at $5,499.
New Apple Desktop Pricing
| Product | New Chip Options | Starting Price |
|---|---|---|
| Mac mini | M6 / M5 Pro | $899 |
| Mac Studio | M5 Max / M5 Ultra | $2,499 / $5,499 |
The new desktop systems are available for preorder, with deliveries and store availability scheduled to begin September 22.
Apple’s Strategy Is Shifting Toward On-Device AI
The chip launches arrive at a time when Apple is facing intense competition in AI. While the company has invested heavily in on-device processing, it has lagged some competitors in developing its own frontier AI models and cloud infrastructure.
Apple’s approach is instead to use its control over hardware and software to make local AI increasingly capable. The M6 and M5 Ultra strengthen that strategy by providing developers with more CPU, GPU and Neural Engine capacity.
The M5 Ultra’s 512GB memory ceiling is particularly notable for AI developers because large models can require enormous amounts of memory. Apple is also highlighting the ability to connect multiple Mac Studio systems using high-speed connectivity for workloads that exceed the capacity of an individual machine.
Why The 2nm Transition Matters
Moving the M6 to a 2nm manufacturing process is an important technical milestone for Apple. Smaller process nodes can enable greater transistor density and improved power efficiency, although the real-world benefits depend on architecture, workload and system design.
Apple is using the process improvement not simply to make the chip smaller, but to expand several parts of the processor simultaneously. The M6 adds CPU and GPU resources, increases memory bandwidth and expands its AI capabilities.
That combination reflects Apple’s broader strategy of treating the CPU, GPU, Neural Engine and memory system as a unified platform rather than independent components.
The Bigger Picture
Apple’s M5 Ultra and M6 launches demonstrate how the company’s custom-silicon strategy is increasingly centered on AI as well as traditional computing. M5 Ultra pushes desktop performance into territory aimed at professional creators and AI developers, while M6 brings stronger local AI capabilities to a lower-cost Mac platform.
The more important shift is Apple’s growing emphasis on local AI computing. As developers experiment with AI agents and larger language models, hardware that can process models quickly without constantly relying on cloud infrastructure could become an increasingly important differentiator.
Looking Ahead
Apple’s immediate challenge will be translating its chip-level improvements into compelling applications and developer adoption. The M6 gives mainstream Mac users and developers more local AI capacity, while M5 Ultra provides a much larger memory and compute envelope for professional AI and creative workloads. The success of that strategy will depend partly on whether Apple’s software tools and AI frameworks can fully exploit the additional hardware.
The broader competition will also intensify as chipmakers and technology companies develop increasingly specialized AI processors. Apple’s advantage remains its ability to control the silicon, operating system and developer ecosystem together. If that integration allows more sophisticated AI workloads to run efficiently on Macs, the M6 and M5 Ultra could become important building blocks in Apple’s longer-term on-device AI strategy.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
