On August 11, 2026, IBM and Together AI officially signed a multi-year cooperation agreement worth $240 million. The two parties will deploy a large-scale AI computing cluster on IBM Cloud, dedicated to inference services for open-source AI models. The cluster is expected to go live in Q1 2027.

Strategic Collaboration and Technical Architecture

This collaboration brings together the core strengths of three parties. IBM provides enterprise-grade cloud infrastructure and hybrid cloud governance capabilities, Together AI contributes leading inference optimization technology and open-source model operation expertise, and NVIDIA supplies the underlying computing power through its HGX B300 systems. Vipul Ved Prakash, CEO of Together AI, noted that enterprises desire cutting-edge model performance but are reluctant to bear the high costs of proprietary models, and the infrastructure jointly built by IBM and NVIDIA precisely provides scalable, highly reliable foundational support for this need.

The cluster adopts NVIDIA HGX B300 systems, integrating the next-generation Blackwell Ultra processors. A single system can deliver up to 72 petaFLOPS of FP8 computing power and 144 petaFLOPS of FP4 computing power, equipped with over 2TB of HBM3e high-bandwidth memory. The cluster achieves high-speed interconnection through NVIDIA Spectrum-X Ethernet networking, with per-GPU network bandwidth reaching 800Gb/s, effectively reducing communication bottlenecks in distributed inference.

Market Significance

Together AI currently processes approximately 400 trillion tokens of inference requests per month, and this $240 million compute commitment will significantly expand its service capacity. Notably, just weeks before the agreement was signed, Together AI completed an $800 million Series C funding round at an $8.3 billion valuation, with NVIDIA also participating.

CONEVO Integrated Circuit Supplier

CONEVO is a global IC distributor specializing in FPGA, MCU, DSP, memory and data converters. Serving OEM and EMS partners worldwide with quality sourcing and agile supply chain solutions, it covers aerospace, automotive, medical, and industrial markets. Daily Hot Picks:

1. ADS7853IPWR — TI 14-bit dual-channel 1MSPS simultaneous-sampling SAR ADC with SPI interface, optimized for motor control and power-quality monitoring.

2. LT1395CS6#TRPBF — ADI 400MHz current-feedback amplifier with 800V/μs slew rate and shutdown mode, ideal for high-speed video and portable RF front-ends.

3. GD5F2GQ4UEYIGR — GigaDevice 2Gbit SPI NAND Flash featuring Quad I/O up to 120MHz, designed for reliable embedded storage in IoT and consumer devices.

Website: www.conevoelec.com

Email: info@conevoelec.com