DCD Magazine #62 - The coming wave
Will compute at sea sink or swim?
News - IT Hardware & Semiconductors, Sebastian Moss, DatacenterDynamics
The latest news from the AI, data center, telco, chip, and cloud sector
DCD Magazine #62 out now
-
Nvidia revenue doubles over last year, as company forecasts record growth
Says it is making more money per gigawatt of data center than ever before
-
OpenAI details Jalapeño AI chip, with 700W TDP
At Hot Chips, company touts its first foray into hardware
-
IBM mainframe chip to run Arm and Z workloads on the same cores
Two worlds on one chip
-
Groq raises $350m, with Nvidia participation "planned"
Company value nearly halved following Nvidia license deal
-
SpaceX earnings: 2GW of compute by year end, aiming for 10-20GW by end of 2027
But Elon Musk's businesses have a habit of missing targets
-
Nvidia to invest $5bn in Safe Superintelligence, Sutskever firm to substantially increase GPU compute
Ilya's AI lab had been using Google TPUs
-
AMD partners with big chip co. Cerebras for ultra-low-latency and high throughput AI inference system
Helios and Cerebras go up against Nvidia's NVL72 and Groq LPX
-
OpenAI to deploy AMD Helios rackscale system as first part of up to 6GW partnership
Company has already tested the beefy AI rack
-
AMD launches Instinct MI400 Series GPUs for AI workloads
Targets both frontier AI and supercomputing
-
AMD officially launches Helios rackscale system, with MI455X GPUs, Venice Epyc CPUs, and Pensando
Following ZT acquisition, company looks to compete with Nvidia's system-level approach
-
SK Hynix raises $26.5bn in blockbuster US listing
The largest ever listing by a foreign firm
-
SambaNova raises $1bn in Series F round, AI compute company valued at $11bn
JPMorganChase to deploy RDUs for on-prem inference
-
OpenAI and Broadcom unveil 'Jalapeño' Intelligence Processor for LLM inference
"A blank-slate design for modern LLM inference, not a general-purpose accelerator adapted from earlier AI workloads"
-
Anthropic signs supply agreement with Micron, following investment
Micron will also use Claude
-
LineShine: All-CPU Chinese supercomputer named world's most powerful
Huawei gear used in 2.198 exaflops supercomputer
-
DARPA Low Resource Computing program aims to serve the Edge with non-traditional micro-computation
Compares the power of musical greeting card silicon to early mainframes
-
OpenAI's custom chip program lead Clive Chan joins Anthropic
Had been working on Broadcom AI chip, expected later this year
-
OpenAI president Greg Brockman owns stakes in Cerebras, CoreWeave, Stripe, and Helion
All companies with which OpenAI has entered into business deals
-
DARPA Biological Technologies Office to lease 50 Nvidia HGX H100s for NODES program
Newer hardware can also be proposed, but must be ready in a month
-
DARPA Smash: Agency hopes to recover all elements from ore, render China’s rare earth supply chain stranglehold obsolete
Would also solve copper shortages and other supply challenges
-
Anant Nivarti rejoins Tesla to lead silicon engineering, including AI6 and Dojo 3
After Dojo team was effectively disbanded, now it's back and targeting space
-
Nvidia LPU-based LPX rack to consume up to 160kW, be fully liquid cooled
The same power consumption as its neighboring Vera Rubin rack
-
-
Nvidia offers DGX Cloud to AI foundation model labs for new Nemotron Coalition
Locking in open source support
-
Nvidia Vera CPU enters full production, pitched at agentic AI workloads
New rack to feature 256 liquid-cooled Vera CPUs
-
Nvidia announces Groq 3 LPU AI inference chip, plans 256-LPU rack
LPX rack will be liquid cooled, focused on "extreme low latency token generation"
-
Nvidia builds out LPU chip team following $20bn Groq acquihire, announcement expected for GTC
Multiple hires for "Nvidia's new class of LPU technology," despite company's Language Processing Unit yet to be officially confirmed
-
