$70.95
24h
-0.84%
7 Jul, 02:24
Meituan identified as AI company with supply chain understanding
Bullish+592.98%
AI Summary
Author argues AI infrastructure has entered phase 2 where the entire supply chain beyond GPUs matters, citing Samsung's earnings surge as evidence; identifies DS and Meituan as AI companies understanding model, engineering, and supply chain.
The Revelation of Samsung Becoming the World's Most Profitable Company!
Samsung Electronics this second quarter is indeed a bit exaggerated. According to preliminary results, Samsung expects operating profit for Q2 2026 to reach 89.4 trillion won, approximately 58.4 billion USD, a surge of about 19 times year-over-year; revenue for the same period is expected to reach 171 trillion won, also significantly higher than market expectations.
What's more interesting is that Samsung's operating profit for this quarter exceeded Nvidia's operating profit of approximately 53.5 billion USD in the previous quarter. In other words, on this quarter's income statement, Samsung has temporarily surpassed Nvidia to become one of the most profitable companies in the world.
This also means that AI infrastructure has entered the second phase.
In the past two years, everyone was focused on GPUs. Whoever has GPUs, whoever has training clusters, whoever has inference clusters, holds the say in computing power. Therefore, H200, B200, GB300 have become the hard currency of the AI industry.
And now, the entire storage price system has been revalued by AI demand.
Everyone has realized that what truly determines the ceiling of AI infrastructure is a complete system: GPU, HBM, DRAM, NAND, networking, packaging, cooling, data centers, software stacks, scheduling systems, down to components, PCB, transformers, PDU, and so on.
Samsung's profit surge today is actually a reminder to the AI industry: computing power is not just GPUs, computing power is the supply chain.
The cost curve of AI infrastructure is being repriced.
Previously, practitioners thought that as long as GPU prices stabilized, AI inference costs would continue to decline.
But now looking at it, things are far from that simple: ASICs might be cheaper than GPUs, but HBM has risen; the unit price of computing power might drop, but memory and storage have risen; model architectures might be more efficient, but user usage volume has exploded exponentially; inference optimization is improving, but context windows, multi-Agent calls, and multimodal inputs are eating up all optimization dividends.
This is the most real contradiction in the AI industry going forward: technology is reducing costs, demand is swallowing the cost reductions.
So Samsung's financial report is more like a new bill submitted by the entire AI industry chain to the market:
In the past, everyone paid for GPUs.
Now, everyone must pay for GPU + Memory + Storage.
In the future, everyone will inevitably have to pay for electricity, networking, scheduling, and delivery capabilities.
This is also why I have always believed that AI companies must not and cannot only tell model stories. AI companies that can truly survive must ultimately understand three things:
First, understand models;
Second, understand engineering;
Third, understand the supply chain.
So, friends, here comes the question, among these model companies, who fits this characteristic?
Hahaha, personally I think DS and Meituan are both quite interesting.