Home News

company news about Majestic Labs wants to solve memory-bound GPU problem

All Products

Rack Storage Server
(179)

Huawei Fusion Server
(31)

Dell Poweredge Server
(59)

H3C Server
(31)

Datacom Switches
(96)

WLAN Device
(21)

Smart Wireless Router
(17)

Hard Drive HDD
(78)

Internal Hard Drive SSD
(16)

Geforce Graphic Card
(27)

INTEL CPU Processor
(20)

Server Memory RAM
(6)

Refurbished Storage Server
(6)

SFP Transceiver Module
(4)

Fibre Channel Switch
(125)

Certification

Customer Reviews

The sales staff of Beijing Qianxing Jietong Technology Co.,Ltd are very professional and patient. They can provide quotations quickly. The quality and packaging of the products are also very good. Our cooperation is very smooth.

—— 《Festfing DV》LLC

When I was looking for intel CPU and Toshiba SSD urgently, Sandy from Beijing Qianxing Jietong Technology Co., Ltd gave me a lot of help and got me the products I needed quickly. I really appreciate her.

—— Kitty Yen

Sandy of Beijing Qianxing Jietong Technology Co.,Ltd is a very careful salesman, who can remind me of configuration errors in time when I buy a server. The engineers are also very professional and can quickly complete the testing process.

—— Strelkin Mikhail Vladimirovich

We are very happy with our experience working with Beijing Qianxing Jietong. The product quality is excellent, and delivery is always on time. Their sales team is professional, patient, and very helpful with all our questions. We truly appreciate their support and look forward to a long-term partnership. Highly recommended!

—— Ahmad Navid

Quality： “Great experience with my supplier. The MikroTik RB3011 was already used, but it was in very good condition and everything works perfectly. Communication was fast and smooth, and all my concerns were addressed quickly. Very reliable supplier—highly recommended.”

—— Geran Colesio

I'm Online Chat Now

Company News

Majestic Labs wants to solve memory-bound GPU problem

Startup Majestic Labs aims to resolve bottlenecks stemming from LLM KV caching by eliminating AI inference’s reliance on high-bandwidth memory.

Its Prometheus server design replaces costly GPUs with proprietary Ignite AI Processing Units (AIUs): monolithic hybrid dies integrating datacenter-grade ARM cores plus RISC-V vector and tensor engines optimized for LLMs. All AIUs operate within a single shared memory pool built on LPDDR6 DRAM. One server hosts 1 to 12 AIUs accessing 8–128 TB of disaggregated unified memory with a collective 25.6 TB/s bandwidth ceiling.

By contrast, an NVIDIA DGX B300 AI server fitted with 8 Blackwell GPUs only delivers 2.3 TB of HBM3e GPU memory plus up to 4 TB DDR5 system memory, alongside 14.4 TB/s peak low-latency interconnect bandwidth. The Prometheus platform delivers over 50× the NVIDIA rig’s GPU memory capacity and 1.7× its bandwidth.

Majestic’s co-founders argue the GPU-HBM pairing pairs premium, high-cost compute with severely constrained memory capacity. Limits arise from restricted GPU die edge connectivity, short interposer cabling lengths, and practical barriers to scaling HBM past 12 layers.

They label GPU-HBM architecture a memory-bound dead end. Instead, the firm proposes cost-efficient processors linked to vastly expanded memory pools via scalable custom Memory Aggregation Chiplets (MACs) connected by miniature copper cables up to one meter long. MACs sit adjacent to onboard memory chips, with each chiplet feeding numerous DRAM dies.

The server’s unified memory space remains fully contiguous and coherent. AIUs communicate through shared memory in a mesh topology tied to MACs, whose total count is undisclosed.

Simple math illustrates the memory scale: standard 2 GB LPDDR6 dies mean a 128 TB Prometheus server requires up to 64,000 individual LPDDR6 dies, necessitating dozens, likely hundreds, of MAC chiplets per unit.

A standard 40U rack can hold up to four Prometheus servers, each populated with 12 AIUs. The full rack draws 120 kW and relies on cold-plate liquid cooling. Majestic claims one such rack delivers the same high-speed memory as 25 NVIDIA NVL72 Vera Rubin racks at a fraction of the power draw, unlocking large-scale AI workloads for firms unable to afford hyperscaler-grade infrastructure. Its design delivers up to 1,000× more memory per processor.

Removing memory bottlenecks enables efficient hosting of state-of-the-art giant frontier models with ultra-long context windows. It optimizes agentic AI, reasoning, graph neural networks, tabular models, video generation and emerging AI architectures, supporting 100× more users per rack while slashing power consumption. Clients can run multi-trillion-parameter models, massive context windows, agent systems and MoE frameworks on a single server with drastically lower power and capital costs.

Majestic forecasts its hardware will cost 10 to 50 times less than equivalent-performance GPU servers upon launch next year, alongside superior power efficiency. The Prometheus server complies with OCP standards and natively supports PyTorch, vLLM and OpenAI’s Triton inference frameworks for seamless migration of existing AI models.

Footnote

Majestic Labs was founded in Tel Aviv in 2023 by three ex-Google and Meta chip design, manufacturing and shipping veterans: CEO Ofer Shacham, President Sha Rabii and COO Masumi Reynders. The firm employs roughly 40 staff across Tel Aviv and a Los Angeles office, securing a $100 million Series A funding round in late 2025. It licenses third-party accelerator IP and develops a custom core for its Ignite AIU. Majestic reports substantial pre-orders from enterprise, private cloud and hyperscaler target customers.

Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!

Pub Time : 2026-07-24 13:49:55 >> News list

Contact Details

Beijing Qianxing Jietong Technology Co., Ltd.

Contact Person: Ms. Sandy Yang

Tel: 13426366826

company news about Majestic Labs wants to solve memory-bound GPU problem

Rack Storage Server

Huawei Fusion Server

Dell Poweredge Server

H3C Server

Datacom Switches

WLAN Device

Smart Wireless Router

Hard Drive HDD

Internal Hard Drive SSD

Geforce Graphic Card

INTEL CPU Processor

Server Memory RAM

Refurbished Storage Server

SFP Transceiver Module

Fibre Channel Switch

Footnote

Rack Storage Server

12 Bays 1U Rackmount Server Lenovo ThinkSystem SR630 Rack Server

ThinkSystem SR250 V2 4SFF Rack Storage Server Intel Xeon E-2378G Processor

Intel C621A Rack Storage Server Inspur NF5180M6 1U Rack Mount Server

Huawei Fusion Server

FusionServer 5288 V6 4U Rack Server 32 DDR4 DIMMs 44 3.5 Inch Hard Disks

Ultra High Density Huawei Fusion Server 1U Network Storage Server 1288H V5

New Gen OceanStor 5310 Huawei Rack Server Hybrid Flash Storage