@FinanceYF5: AI chip startups are now attacking the 'data movement' bottleneck from multiple angles, aiming to challenge Nvidia's dominance. Deedy compiled a list from July 2026 — let's go through them one by one.
Summary
This article discusses how next-generation AI chip startups are attacking the data movement aspect from different angles in an attempt to challenge Nvidia's dominance, citing a list from July 2026.
View Cached Full Text
Cached at: 07/27/26, 09:58 PM
Now, startups making next-generation AI chips are all attacking the “data shuttling” link from different angles, aiming to shake Nvidia’s dominance.
Deedy has compiled a list for July 2026, let’s take a look one by one👇 https://t.co/2jCYlhAIhB
– Eliminate DRAM (Groq) – Eliminate inter-chip interconnects (Cerebras) – Eliminate separation of computation and memory (d-Matrix) – Eliminate server-centric computing architecture (Majestic) – Abandon generality (Etched, Taalas, MatX) – Or get rid of the $400 million lithography machine (Substrate)
This list does not include companies in the interconnect and alternative substrate fields; some are added here.
Source:
1/ Yesterday someone hiked for 4 hours in the redwood forest in Northern California, working with ChatGPT Voice via AirPods the entire time.
The work done in those 4 hours is more than what he normally does in 8 hours at his desk.
Similar Articles
@snowboat84: https://x.com/snowboat84/status/2061962883651731602
This article is the first part of the AI Engineering Panorama series. From a historical perspective, it reviews the evolution of GPUs from gaming graphics cards to AI accelerators, the bold bet of CUDA, the independent path of Google's TPU, and why NVIDIA ultimately prevailed. It also provides a detailed analysis of the underlying logic of AI infrastructure such as chips, supply chain, networking, and power.
@FinanceYF5: 1/ DeepSeek is playing a trillion-dollar game. It doesn't do programming packages, doesn't do multimodal, and insists on open source — looks like self-sabotage. The truth is: it's not aiming for a few hundred million dollars in business, but to support a $10 trillion Chinese AI hardware ecosystem.
DeepSeek doesn't do programming packages, doesn't do multimodal, and insists on open source — seemingly sabotaging itself, but in fact aims to promote a $10 trillion Chinese AI hardware ecosystem.
@GoSailGlobal: https://x.com/GoSailGlobal/status/2068243415070826738
GPU utilization in the AI industry is generally below 50%. Former a16z partner Anjney Midha founded AMP, aiming to dispatch computing power like electricity to improve utilization efficiency. The article also discusses Anthropic's success strategy, DeepMind's paper hoarding problem, and the correct approach for non-NVIDIA chips.
@Saccc_c: Follow Lao Huang, easy money in hand. Brothers who didn't make it in the first half of the year should hurry up! Lao Huang's call ability is superb. Yesterday, the industry chain background image went viral, and AI cloud sector companies like $NBIS exploded. Now Marvell is surging in after-hours trading. Who's next? I've sorted out the US stock wealth codes announced at NVIDIA GTC: 1…
This post sorts out listed companies mentioned at the NVIDIA GTC conference in areas such as AI computing cloud (e.g., $CRWV, $NBIS, $IREN), AI factory software platform ($IBM), AI data center design and construction ($CDNS, $J), power distribution and liquid cooling temperature control ($VRT, $ETN, $GEV), and AI data center infrastructure ($EQIX, $DLR) as investment references.
@FinanceYF5: SK hynix and NVIDIA have signed a multi-year cooperation agreement. Memory chips cannot wait for GPU design to be completed; advanced DRAM requires years of joint design, manufacturing planning, and capital investment. AI supercomputers, personal AI PCs, Jetson robot platforms. Using NVIDIA tools to create a digital twin of the chip factory, first running all tests in the virtual factory before the actual production line...
SK hynix and NVIDIA have signed a multi-year cooperation agreement covering joint design and manufacturing planning of advanced DRAM, as well as AI infrastructure (supercomputers, AI PCs, Jetson) and digital twin factory technology, marking that the AI infrastructure competition has reached the memory level.