Decoding NVIDIA’s and Poolside’s latest deal: NVIDIA pays $6 billion to obtain a non-exclusive license for Poolside’s “Model Factory” technology—core software for Poolside’s construction of the Laguna series open-source code models. At the same time, NVIDIA has extended invitations to more than 100 Poolside engineers, who will join NVIDIA’s Nemotron open-weight model project. Since it launched in 2023, the project has been developing larger, more advanced versions; reports suggest the parameter scale could reach the trillion level. There is also roughly a $1 billion equity investment to go along with it. This is a signature move from NVIDIA’s “plow with a hoe” to “dig for gold,” as Huang Renxun uses it to dive deeper into the model layer and build full-stack “AI factory” capabilities tightly integrated with NVIDIA’s ecosystem.
Four key takeaways:
First NVIDIA aims to recreate one of the world’s strongest open-weight models, benchmarking China’s open-source models while directly challenging leading U.S. closed-source companies such as OpenAI and Anthropic. Open-weight models are cheaper to run and easier to customize.
Second Expanding from selling GPU hardware to the software and model layers strengthens NVIDIA’s control across the entire upstream and downstream ecosystem.
Third The moat deepens, but competition with customers also intensifies: improvements in open-weight model capabilities may drive GPU demand, but Nemotron directly competes with large customers’ closed models, which could strain relationships.
Fourth The trend can be summarized as large-model providers developing their own chips, and chipmakers developing their own models—competition is further escalating.
When Huang Renxun publicly voiced support for open source at the end of July, it was effectively laying the groundwork for NVIDIA’s own open-source models. This deal is a crucial step for NVIDIA to push harder in the open-source race between China and the United States. If the upgraded Nemotron’s performance can catch up, it will further benefit cloud providers and software companies that control customer workflow data.
On August 26 at 20:30, the July PCE data will be released. If the reading comes in hotter than expected, expectations for the Federal Reserve’s September policy path are likely to be affected, and fluctuations in U.S. Treasury yields may move in tandem with global risk assets. In addition, the revised estimate of the U.S. real GDP for Q2 will be released on August 26 at 20:30. On a month-over-month basis, investors expect the Fed’s preferred inflation gauge—the core PCE price index—to rise by 0.2%. If the increase is larger, it could lead investors to reconsider expectations that the Fed will stay on hold in September, potentially opening the door for a pullback in gold. Conversely, if the monthly core PCE inflation data is weaker than expected, it may help gold move higher further.
Then on Thursday, August 27, traders will focus on the weekly initial jobless claims report. At 22:00 on Friday, the preliminary annual benchmark revision to nonfarm payrolls data will also be released, along with the final reading of the University of Michigan’s August consumer sentiment index.
The preliminary annual benchmark revision to nonfarm employment data will indicate whether previously reported U.S. employment figures were overestimated or underestimated. After recent signs of softer hiring, this revision may become even more important. If it suggests a significant weakening in the labor market, it could affect expectations for Fed policy.
Decoding NVIDIA’s and Poolside’s latest deal: NVIDIA pays $6 billion to obtain a non-exclusive license for Poolside’s “Model Factory” technology—core software for Poolside’s construction of the Laguna series open-source code models. At the same time, NVIDIA has extended invitations to more than 100 Poolside engineers, who will join NVIDIA’s Nemotron open-weight model project. Since it launched in 2023, the project has been developing larger, more advanced versions; reports suggest the parameter scale could reach the trillion level. There is also roughly a $1 billion equity investment to go along with it. This is a signature move from NVIDIA’s “plow with a hoe” to “dig for gold,” as Huang Renxun uses it to dive deeper into the model layer and build full-stack “AI factory” capabilities tightly integrated with NVIDIA’s ecosystem.
Four key takeaways:
First NVIDIA aims to recreate one of the world’s strongest open-weight models, benchmarking China’s open-source models while directly challenging leading U.S. closed-source companies such as OpenAI and Anthropic. Open-weight models are cheaper to run and easier to customize.
Second Expanding from selling GPU hardware to the software and model layers strengthens NVIDIA’s control across the entire upstream and downstream ecosystem.
Third The moat deepens, but competition with customers also intensifies: improvements in open-weight model capabilities may drive GPU demand, but Nemotron directly competes with large customers’ closed models, which could strain relationships.
Fourth The trend can be summarized as large-model providers developing their own chips, and chipmakers developing their own models—competition is further escalating.
When Huang Renxun publicly voiced support for open source at the end of July, it was effectively laying the groundwork for NVIDIA’s own open-source models. This deal is a crucial step for NVIDIA to push harder in the open-source race between China and the United States. If the upgraded Nemotron’s performance can catch up, it will further benefit cloud providers and software companies that control customer workflow data.
Decoding NVIDIA’s and Poolside’s latest deal: NVIDIA pays $6 billion to obtain a non-exclusive license for Poolside’s “Model Factory” technology—core software for Poolside’s construction of the Laguna series open-source code models. At the same time, NVIDIA has extended invitations to more than 100 Poolside engineers, who will join NVIDIA’s Nemotron open-weight model project. Since it launched in 2023, the project has been developing larger, more advanced versions; reports suggest the parameter scale could reach the trillion level. There is also roughly a $1 billion equity investment to go along with it. This is a signature move from NVIDIA’s “plow with a hoe” to “dig for gold,” as Huang Renxun uses it to dive deeper into the model layer and build full-stack “AI factory” capabilities tightly integrated with NVIDIA’s ecosystem.
Four key takeaways:
First NVIDIA aims to recreate one of the world’s strongest open-weight models, benchmarking China’s open-source models while directly challenging leading U.S. closed-source companies such as OpenAI and Anthropic. Open-weight models are cheaper to run and easier to customize.
Second Expanding from selling GPU hardware to the software and model layers strengthens NVIDIA’s control across the entire upstream and downstream ecosystem.
Third The moat deepens, but competition with customers also intensifies: improvements in open-weight model capabilities may drive GPU demand, but Nemotron directly competes with large customers’ closed models, which could strain relationships.
Fourth The trend can be summarized as large-model providers developing their own chips, and chipmakers developing their own models—competition is further escalating.
When Huang Renxun publicly voiced support for open source at the end of July, it was effectively laying the groundwork for NVIDIA’s own open-source models. This deal is a crucial step for NVIDIA to push harder in the open-source race between China and the United States. If the upgraded Nemotron’s performance can catch up, it will further benefit cloud providers and software companies that control customer workflow data.
Decoding NVIDIA’s and Poolside’s latest deal: NVIDIA pays $6 billion to obtain a non-exclusive license for Poolside’s “Model Factory” technology—core software for Poolside’s construction of the Laguna series open-source code models. At the same time, NVIDIA has extended invitations to more than 100 Poolside engineers, who will join NVIDIA’s Nemotron open-weight model project. Since it launched in 2023, the project has been developing larger, more advanced versions; reports suggest the parameter scale could reach the trillion level. There is also roughly a $1 billion equity investment to go along with it. This is a signature move from NVIDIA’s “plow with a hoe” to “dig for gold,” as Huang Renxun uses it to dive deeper into the model layer and build full-stack “AI factory” capabilities tightly integrated with NVIDIA’s ecosystem.
Four key takeaways:
First NVIDIA aims to recreate one of the world’s strongest open-weight models, benchmarking China’s open-source models while directly challenging leading U.S. closed-source companies such as OpenAI and Anthropic. Open-weight models are cheaper to run and easier to customize.
Second Expanding from selling GPU hardware to the software and model layers strengthens NVIDIA’s control across the entire upstream and downstream ecosystem.
Third The moat deepens, but competition with customers also intensifies: improvements in open-weight model capabilities may drive GPU demand, but Nemotron directly competes with large customers’ closed models, which could strain relationships.
Fourth The trend can be summarized as large-model providers developing their own chips, and chipmakers developing their own models—competition is further escalating.
When Huang Renxun publicly voiced support for open source at the end of July, it was effectively laying the groundwork for NVIDIA’s own open-source models. This deal is a crucial step for NVIDIA to push harder in the open-source race between China and the United States. If the upgraded Nemotron’s performance can catch up, it will further benefit cloud providers and software companies that control customer workflow data.
Market conditions change in the blink of an eye, with trending themes cycling through one after another. Don’t let emotions sweep you up, and don’t chase highs blindly—view every rise and fall rationally. Stay calm, hold your position, and wait for the wind to come. Wishing all you family members great returns—may everything you hope for come true 🚀
📢 Alert! “Niu Lai”’s issuing address snatched $155,000 in fees—are you still rushing in?
GMGN data confirms it—just 20 hours ago, the “Niu Lai” address launched yet another new project called “Niu Lai Life,” and this is already its 12th project! Cumulative fees have reached 224.17 BNB, worth about $155,000.
⚠️ This isn’t “Niu Lai”—it’s “Shave Lai”! For these high-frequency issuing addresses, the usual playbook is mass harvesting and quick in/quick out, with zero real ecosystem support. It’s a classic risk bomb. Market funds are repeatedly drained, retail trust is repeatedly crushed—sell-pressure signals are already flashing red.
💡 Retail survival guide:
· Avoid all related projects tied to this address—don’t be a “bag holder.” · Keep a close eye on where its funds flow—often it’s an early warning sign of risk. · Don’t chase pumps, don’t FOMO—protecting your principal is the real strategy.
The hotter the market, the calmer you must be. The “opportunities” you see may be traps set up by others. Share the warning to help more people avoid the pit!👇
Lucy’s curve soars like the soaring dragon, with its head held high, clouds surge toward the heavens. A strong lineup in the Bright Community— one flight lifts straight up into outer space.
LUCIC’s Light rises from the palm, spilling over the starlit city and the sea. May the distant shores you journey toward be as gentle as poetry, shining brightly.
The momentum of U.S. economic growth is accelerating. S&P Global’s U.S. Composite PMI rose by 1.5 points in August to 56.0, the highest level since April 2022 and the third consecutive month of gains. At the same time, the Services PMI increased by 2.2 points to 56.8, the highest level since March 2022. The Manufacturing PMI fell by 0.7 points to 53.9, the lowest level in the past five months, but it remains in the expansion range.
Meanwhile, U.S. firms’ hiring activity is led by the services sector, with headcount increasing at the fastest pace since January 2025. These data suggest that the United States’ year-on-year GDP growth rate in Q3 2026 will reach +3.0%, compared with +1.5% in Q2. Artificial intelligence is driving a historic wave of growth.
Anthropic’s In-House Chip Bet: Poaches Google’s TPU Business Founder—A “Decouple from Nvidia” Gamble on the Eve of a $200B IPO
Anthropic is shifting its strategic focus from “battling for compute power” to “building compute capacity.” The company has hired Amir Salek, the founder of Google’s custom chip business, to assemble an in-house chip team, while also collaborating with multiple partners including Fractile, Google, Broadcom, and Samsung to target approximately 3.5GW of TPU capacity by 2027. This “in-house R&D + diversified procurement” dual-track strategy is designed to break away from reliance on Nvidia and reduce inference costs. During the critical window to launch an IPO at a $200 billion valuation, Anthropic is shifting its strategic focus from “battling for compute power” to “building compute capacity,” seeking to directly take control of the foundational infrastructure that supports its explosive growth.