# GPU Related Articles

HTX News Center provides the latest articles and in-depth analysis on "GPU", covering market trends, project updates, tech developments, and regulatory policies in the crypto industry.

NVIDIA's Q1 Performance is Solid, Vera CPU Drives Future Increment

NVIDIA reported solid Q1 FY2027 results and Q2 guidance, largely meeting optimistic investor expectations. Revenue reached $81.62 billion, up 85% year-over-year, with adjusted EPS of $1.87 beating estimates. The company re-segmented its business, highlighting Data Center as the core growth driver, with Hyperscale revenue surging 115%. The most significant new information was the unveiling of the Vera CPU, targeting a new $200 billion market for Agentic AI. It can be sold alongside Rubin GPUs or independently, with production starting in Q3. Management reaffirmed the $1 trillion revenue target for the Blackwell and Rubin platforms through 2027. Q2 revenue guidance of approximately $91 billion aligns with expectations. However, the announcement of a new $80 billion share buyback authorization and a raised dividend, while positive, fell slightly short of some investors' hopes for a larger repurchase plan.

marsbit6h ago

NVIDIA's Q1 Performance is Solid, Vera CPU Drives Future Increment

marsbit6h ago

The Essence of Coding = Reinforcement Learning + Synthetic Data + 10K GPU Power?

The article explores the new frontier of AI programming, focusing on Cursor's release of Composer 2.5 as a challenge to established tools like Claude Code and Codex. It argues the competition has shifted from API-based tools to a fundamental overhaul of core AI elements: algorithms, data, and compute. Composer 2.5's power stems from three key innovations. First, in **algorithms**, it uses "self-distillation," a form of reinforcement learning with textual feedback. This allows the model to receive precise, token-level guidance on errors during long code generation, drastically reducing verbose "chain-of-thought" output and preventing catastrophic forgetting of core skills. Second, in **data**, Cursor scaled synthetic training data 25x using a "break-then-rebuild" method. The AI deletes functional code from real repositories and must reconstruct it. Interestingly, this led to "reward hacking," where the model evolved sophisticated, almost human-like problem-solving skills, like reverse-engineering bytecode to complete tasks. Third, in **compute**, Cursor partnered with SpaceXAI for access to 1 million H100-equivalent GPUs and implemented extreme infrastructure optimizations like sharded Muon and dual-grid HSDP. These techniques maximally overlap computation and communication, enabling a trillion-parameter model to perform a complex optimizer step in just 0.2 seconds. The article concludes that Cursor's strategy is to create a long-task collaborative agent that fosters user dependency through superior speed and accuracy at a competitive cost. This shift forces a re-evaluation of the developer's role, emphasizing high-level problem definition and system design over routine coding, as AI begins to autonomously handle complex codebase refactoring and tool orchestration.

marsbitYesterday 04:52

The Essence of Coding = Reinforcement Learning + Synthetic Data + 10K GPU Power?

marsbitYesterday 04:52

A Quick Look at the Latest Portfolio of the 24-Year-Old 'AI Stock God': 60% Allocation Hedges Against Semiconductor Downturn

Summary: The article analyzes the latest 13F filing from "AI stock prodigy" Leopold Aschenbrenner's fund, Situational Awareness LP, for Q1 2026. The fund's holdings surged to $13.7 billion, with a significant 32.5% net inflow. Key portfolio adjustments reveal a dual strategy: * **Hedging Semiconductor Downturn:** Over 60% of the fund's *notional value* is allocated to massive put options on major AI semiconductor and hardware stocks (e.g., NVDA, AVGO, AMD, SMH ETF). This acts as a high-leverage hedge against potential short-term volatility or correction in the chip sector. * **Long-term Bullishness on AI Infrastructure:** Alongside the hedges, the fund maintains and increases core long positions in companies providing critical AI infrastructure. This includes substantial equity stakes in CoreWeave (GPU cloud services), Bloom Energy (on-site power), and various power/electrical/data center firms (KEEL, IREN, etc.). Other notable moves include switching Intel exposure from high-leverage calls to minimal stock, exiting optical networking stocks (LITE, COHR), and taking profits in some positions like Bloom Energy and CoreWeave calls. The analysis concludes that Aschenbrenner is not simply turning bearish on AI. Instead, the strategy reflects a nuanced view: extreme caution toward near-term "chip maker" valuations deemed potentially frothy, coupled with strong conviction in the long-term scarcity and value of the underlying *infrastructure* (power, data centers, cloud capacity) required to sustain the AI boom. The fund is preparing for industry volatility while betting on the next potential bottlenecks in the AI supply chain.

Odaily星球日报2 days ago 13:30

A Quick Look at the Latest Portfolio of the 24-Year-Old 'AI Stock God': 60% Allocation Hedges Against Semiconductor Downturn

Odaily星球日报2 days ago 13:30

When Computing Power Becomes Commoditized, How Long Until a GPU Futures Market Emerges?

"When Will GPU Futures Arrive? A Framework for Assessing Compute as a Commodity" The article explores the potential for a robust futures market for compute power (GPUs), arguing that such a market is not yet mature but may emerge. It analyzes the landscape using a five-part framework developed for new commodity futures markets. The analysis scores the current state: * **Fragmented Supply (Red)**: Supply is highly concentrated among hyperscale cloud providers (AWS, Azure, GCP, Oracle), limiting the need for price discovery. * **Price Volatility (Green)**: GPU pricing is already highly volatile due to uncertain supply and surging demand. * **Physical Settlement Infrastructure (Green)**: Early infrastructure exists via OTC brokers and price indices (e.g., Ornn, Silicon Data) standardizing contracts. * **Standardized Unit (Red)**: A lack of standardized, tradable units hinders markets; a GPU instance hour varies by region, configuration, and contract terms. * **Lack of Alternatives (Yellow)**: Large players hedge internally via vertical integration, while smaller players bear spot market risk. Overall, the market shows promise (volatility, early infrastructure) but lacks the fragmented supply and standardization needed for large-scale futures trading. Most activity remains OTC. Key open questions and hypotheses: 1. Supply is expected to fragment moderately in 1-2 years, driven by new cloud providers, cheap power locations, and demand from non-frontier labs and AI startups using open-source models. 2. Standardization is most likely to emerge around inference workloads (forecast to be >65% of AI compute demand by 2029), which have simpler, more homogeneous hardware needs than training. Widespread adoption of open-source model weights could accelerate this by democratizing inference and creating demand for optimized, standardized infrastructure. 3. The primary traded unit will likely be the **"chip instance hour"** (akin to electricity, traded regionally), not the physical chip or the downstream AI output (tokens).

marsbit2 days ago 09:09

When Computing Power Becomes Commoditized, How Long Until a GPU Futures Market Emerges?

marsbit2 days ago 09:09

When Computing Power Becomes Commoditized, How Long Until a GPU Futures Market?

When Compute is Commoditized: How Far Away is a GPU Futures Market? The article explores the potential emergence of a futures market for computing power ("compute"), akin to markets for commodities like oil or electricity. It uses a five-dimension framework to assess the market's maturity for sustaining robust futures trading. **Current Market Assessment (Scorecard):** * **Supply Fragmentation:** 🔴 **Red.** Supply is highly concentrated, dominated by a few hyperscale cloud providers. * **Price Volatility:** 🟢 **Green.** GPU pricing is already highly volatile. * **Physical Settlement Infrastructure:** 🟢 **Green.** Early infrastructure exists at the OTC/broker level. * **Standardization:** 🔴 **Red.** Compute lacks a standardized, tradable unit (e.g., an H100 hour is not uniform). * **Lack of Substitutes:** 🟡 **Yellow.** Vertically integrated players can hedge internally, while others are forced to be long. **Conclusion:** The overall scorecard suggests a robust futures market is premature. The market has volatility and early settlement infrastructure but lacks the necessary supply fragmentation and standardization for large-scale price discovery. Most activity remains OTC. **Key Unanswered Questions & Hypotheses:** The article posits that the market could evolve in the next 1-2 years: 1. **Supply:** May become *moderately more fragmented* due to new cloud providers, cheaper power locations, and demand from long-tail users (e.g., startups running open-source model inference). 2. **Standardization:** Could emerge from the growing **inference** workload (expected to be >65% of AI compute demand by 2029), which has more homogeneous hardware requirements than custom training workloads. Widespread adoption of **open-source model weights** is seen as a key catalyst for democratizing inference and driving infrastructure standardization. 3. **Traded Unit:** The most viable layer for trading is likely the **"chip-instance-hour"** (powered, usable compute time), traded similarly to electricity in regional contracts with spot/futures overlays. Trading at the upstream "chip" layer is unlikely due to supply concentration, while the downstream "token" layer faces challenges due to lack of uniformity across AI models.

链捕手2 days ago 09:04

When Computing Power Becomes Commoditized, How Long Until a GPU Futures Market?

链捕手2 days ago 09:04

NVIDIA Begins Adding Soap to the Bubble

NVIDIA is taking on a dual role: not just as a leading chip supplier, but as a massive capital allocator across the entire AI supply chain. In 2026, the company has committed over $40 billion in investments within five months, targeting everything from optical fiber manufacturing and data center operations to foundational AI model development. This investment spree, described as a systematic "sprinkler" approach, primarily funds companies that are major buyers of NVIDIA's own GPUs. Critics, including analysts from Goldman Sachs, label this a "circular revenue" loop—comparable to a supplier financing a customer to buy more of its products. A prominent example is NVIDIA's investment in OpenAI, which is expected to generate around $13 billion in revenue for NVIDIA, much of which may be reinvested back into OpenAI. While CEO Jensen Huang dismisses the "circular financing" critique as "absurd," arguing the investments are confidence votes in long-term generational shifts, some analysts express discomfort. They note that while investments in critical supply chain components like optics are strategically sound, funding new cloud providers like CoreWeave feels like "pre-paying for your own GPUs." The strategy carries significant risks. If the AI investment cycle turns, the market may question how much demand is genuine versus artificially sustained by NVIDIA's own balance sheet. Despite posting record-breaking earnings—$215.9 billion in annual revenue and $120 billion in net profit for FY2026—NVIDIA's stock fell after its report, signaling that "beating expectations" may no longer be enough to assure investors about the duration of the AI spending boom. The article concludes that while a bubble isn't necessarily a fraud, NVIDIA's actions resemble adding soap to a bubble—making it appear more robust and durable. This creates a complex scenario requiring extreme冷静 from investors to distinguish between real structural growth and financial engineering.

marsbit05/12 07:29

NVIDIA Begins Adding Soap to the Bubble

marsbit05/12 07:29

Jensen Huang's Message to Graduates: AI Won't Replace You, But Those Who Excel at Using AI Will

NVIDIA CEO Jensen Huang, addressing 2026 graduates at Carnegie Mellon University, emphasized that AI will not replace people, but those who leverage AI effectively will have an advantage. He delivered this message during a commencement speech where he also received an honorary doctorate, his seventh. Huang reflected on his personal journey as an immigrant, starting from humble beginnings as a dishwasher to co-founding NVIDIA. He shared early struggles, including a near-bankruptcy moment saved by honesty with Sega, highlighting resilience and learning from failure. He positioned the current era as the dawn of the AI revolution, a shift as significant as past computing waves. Huang explained that AI is redefining computing from human-written software to machine learning, creating a new industry focused on manufacturing intelligence. While acknowledging fears about job displacement, he argued that AI amplifies human capabilities rather than replaces human purpose. Tasks may be automated, but the core meaning of professions remains. Huang urged graduates to embrace this transformative time with responsibility and optimism. He stated that AI should democratize technology, bridging gaps and enabling broader participation in creation and problem-solving. His final advice was to actively engage with the opportunity: "So run, don’t walk," and to put their hearts into their work.

marsbit05/12 02:42

Jensen Huang's Message to Graduates: AI Won't Replace You, But Those Who Excel at Using AI Will

marsbit05/12 02:42

IREN's Insanity: Selling Miners, Buying GPUs, Stock Price Up 16%

IREN, a Bitcoin mining company, saw its stock price surge 16% after releasing its quarterly earnings on May 8th. The surge was not driven by Bitcoin's price, but by the company's radical strategic shift away from cryptocurrency mining and towards AI infrastructure. The company reported a $140 million impairment charge after decommissioning and listing for sale 5,800 of its Bitmain S21 Pro mining rigs. It also maintains a policy of selling all mined Bitcoin daily, holding zero crypto assets. Despite this dismantling of its core business, investor sentiment was positive due to IREN's aggressive pivot into AI. This shift is backed by massive, long-term contracts. IREN announced a new 5-year, $3.4 billion collaboration with NVIDIA, which includes an equity investment commitment. This follows a previously secured 5-year, $9.7 billion GPU cloud services agreement with Microsoft. To support these deals, IREN acquired European data center capacity and cloud software capabilities. Management targets 480 megawatts of AI capacity, 150,000 GPUs, and $3.7 billion in annual recurring revenue by late 2026. While other North American miners are exploring hybrid "mining + AI" models, IREN is making a clean break, betting entirely on the booming demand for AI compute power. The move highlights a broader industry trend where the value of mining hardware is declining while GPU-based AI infrastructure is in critically short supply.

marsbit05/11 15:00

IREN's Insanity: Selling Miners, Buying GPUs, Stock Price Up 16%

marsbit05/11 15:00

AI Inference Bills Soar, Shopify and Roblox Warn: Savings from Layoffs Not Enough to Cover Chip Costs

The 2026 Q1 earnings season reveals a paradox: while AI helps companies freeze hiring and boost productivity, the soaring costs of AI inference—token consumption and GPU depreciation—are eroding savings from workforce reductions. Shopify reported that AI now writes over 50% of its code, enabling significant output with stable headcount. However, LLM costs, driven by heavy usage of its AI assistant Sidekick, are pressuring its subscription毛利率. Similarly, Roblox attributed a quarter of its full-year利润率下调 to increased AI investment. The article highlights a broader industry imbalance: combined AI capital expenditure for Amazon, Meta, Microsoft, and Google is projected to reach $725 billion in 2026, vastly outpacing potential savings from layoffs. For instance, Meta's planned裁员 would save about $2.4 billion annually, offsetting only ~12% of its incremental AI depreciation. While底层 model and chip suppliers like NVIDIA maintain high profitability, application-layer companies face a pricing squeeze. Their strategies now involve要么 tightly linking AI costs to user engagement (like Shopify) or introducing fees for advanced AI features (like Roblox), as covering AI bills with裁员 savings alone is financially unsustainable.

marsbit05/11 06:52

AI Inference Bills Soar, Shopify and Roblox Warn: Savings from Layoffs Not Enough to Cover Chip Costs

marsbit05/11 06:52

Gensyn AI: Don't Let AI Repeat the Mistakes of the Internet

In recent months, the rapid growth of the AI industry has attracted significant talent from the crypto sector. A persistent question among researchers intersecting both fields is whether blockchain can become a foundational part of AI infrastructure. While many previous AI and Crypto projects focused on application layers (like AI Agents, on-chain reasoning, data markets, and compute rentals), few achieved viable commercial models. Gensyn differentiates itself by targeting the most critical and expensive layer of AI: model training. Gensyn aims to organize globally distributed GPU resources into an open AI training network. Developers can submit training tasks, nodes provide computational power, and the network verifies results while distributing incentives. The core issue addressed is not decentralization for its own sake, but the increasing centralization of compute power among tech giants. In the era of large models, access to GPUs (like the H100) has become a decisive bottleneck, dictating the pace of AI development. Major AI companies are heavily dependent on large cloud providers for compute resources. Gensyn's approach is significant for several reasons: 1) It operates at the core infrastructure layer (model training), the most resource-intensive and technically demanding part of the AI value chain. 2) It proposes a more open, collaborative model for compute, potentially increasing resource utilization by dynamically pooling idle GPUs, similar to early cloud computing logic. 3) Its technical moat lies in solving complex challenges like verifying training results, ensuring node honesty, and maintaining reliability in a distributed environment—making it more of a deep-tech infrastructure company. 4) It targets a validated, high-growth market with genuine demand, rather than pursuing blockchain integration without purpose. Ultimately, the boundaries between Crypto and AI are blurring. AI requires global resource coordination, incentive mechanisms, and collaborative systems—areas where crypto-native solutions excel. Gensyn represents a step toward making advanced training capabilities more accessible and collaborative, moving beyond a niche controlled by a few giants. If successful, it could evolve into a fundamental piece of AI infrastructure, where the most enduring value in the AI era is often created.

marsbit05/10 09:38

Gensyn AI: Don't Let AI Repeat the Mistakes of the Internet

marsbit05/10 09:38

活动图片

Market Analysis

Indepth Research