From July 23 to 24, a multi-line "asset review" surrounding AI unfolded simultaneously among chips, payments, DeFi, and tech stocks: on one side, the decentralized trading aggregation protocol Odos announced it would stop new registrations, transitioning to read-only mode from July 27 to 30 and permanently shutting down after July 30, signaling an exit for the once-hot DeFi aggregation sector; on the other side, Coinbase's incubated x402 protocol began allowing enterprise clients to directly accept USDC payments initiated by AI agents, while traditional payment giant Stripe was reported to be negotiating to acquire the aggregation platform OpenRouter, which consolidates models like GPT-4 and Claude for about $10 billion. Alphabet's stake in Anthropic was estimated to have a book value of about $124 billion, and the valuations of upstream models and payment gateways were re-elevated. At the same time, AMD announced the full production of its rack-level AI system Helios at the Advancing AI conference in San Francisco, projecting a trillion-dollar AI accelerator market by 2030. Intel offered a better-than-expected Q3 revenue outlook, causing its stock price to briefly rise about 13% after hours, while Tesla saw its stock drop over 14% in a single day after second-quarter profits fell short of expectations, with short-sellers' profits exceeding $4.12 billion, making it the most eye-catching example of a pullback among overvalued "tech story stocks." During this window, the AI gold rush clearly preferred to reward shovel sellers and payment infrastructure while starting to squeeze certain crypto protocols and tech companies that tell AI stories but struggle to realize profits.

Odos Shuts Down: DeFi Aggregation Sector Faces Reverse Screening

In the same week that chip manufacturers and payment giants were busy paving the way for the AI gold rush, the decentralized aggregation protocol Odos, which once provided the "optimal path" for on-chain transactions, announced its exit. The team's timetable was clear: starting July 23, all new registrations would be disabled on the application side, between July 27 and 30, the front end and contract interactions would be locked into read-only mode, allowing users to only view historical and current positions without initiating new on-chain exchanges through aggregation routing; after July 30, Odos services would shut down permanently, and no further development, support, or maintenance would be provided. For users accustomed to price comparisons and order splitting across multiple chains and DEXs through aggregators, this meant that a whole routing logic would "disappear" from on-chain, forcing them to revert to individual protocols or other existing aggregators for alternatives.

The departure of Odos cannot simply be viewed as an isolated event of a team "running out of steam;" it seems more like a sample of the current clearing cycle where the DeFi aggregation sector is being reverse screened. Against the backdrop of sustained market pressure and a tightening regulatory environment, protocols lacking scale advantages, brand awareness, or clear paths to profitability are more likely to choose to shut down proactively rather than struggle on in a low-fee, high-risk sector. Notably, it's also a matter of timing: on one hand, institutions like Coinbase announced this week plans to connect on-chain payments with AI agents, Stripe was reported to be trying to acquire a model aggregation platform to access the "computing power entry," and chip manufacturers presented an optimistic growth curve toward 2030 at the conference; on the other hand, foundational tool protocols like Odos silently hit the shutdown button. As developers and capital shift their attention toward "AI shovel sellers" and payment underpinnings, traditional DeFi tools lacking narrative support and a quick way to prove cash flow will be the first to be eliminated in this round of screening, leaving only a few aggregators with scale effects and differentiated value.

Coinbase x402 Launches AI Agent USDC Payments

At almost the same time as Odos chose to shut down, Coinbase's answer was not to create a new aggregator, but to directly rewrite the payment entry. Its incubated x402 protocol was explicitly positioned as the payment underpinnings of the "agent economy:" enterprise clients do not need to rebuild complex on-chain infrastructure to accept USDC payments initiated by AI agents. USDC, as a mainstream US dollar-pegged on-chain settlement asset, has already been adopted by several institutions in scenarios such as cross-border settlements and fund consolidation. Now, Coinbase attempts to elevate this asset from a "human-confirmed" payment tool to a "machine-automated settlement" standard interface.

The underlying imagination is that when AI applications no longer only generate content and make decisions but are authorized to complete payments directly on-chain, the crypto network begins to truly assume the role of the global settlement layer. x402 ties together the enterprise account system, AI application front end, and USDC settlement mainline, essentially creating an "AI + on-chain payment" harness for traditional enterprises: as long as a business accepts USDC deposits, various AI agents at the front end can connect orders, subscriptions, usage-based billing, etc., while the protocol handles signing and accounting logic in the background. For Coinbase, this step is not just about launching a new product; it firmly positions itself in the long-term pathway of "infrastructure for every on-chain dollar payment initiated by AI agents."

Stripe Targets OpenRouter Payment Model Entry

If x402 directly connects on-chain dollar payments to AI agents' "wallets," what Stripe aims to do is to transform the model invocation itself into a receivable scenario. OpenRouter serves as such an entry point: it aggregates various mainstream large models, including GPT-4 and Claude, providing developers with a unified interface that allows applications to connect once to OpenRouter for free switching and orchestration of different models in the backend. Reports indicate that Stripe is negotiating to acquire this model distribution platform for about $10 billion, while OpenRouter was previously valued at around $1.3 billion, highlighting the premium as a reflection of the belief that model aggregation layers are viewed as extensions of payment infrastructure in terms of "who controls AI traffic entry."

Stripe has long provided acquiring capabilities for numerous internet and AI enterprises, but that was just the "settlement endpoint" from an external application perspective. Once it secures OpenRouter, it can extend its reach to the model invocation layer, folding "which model to select, how much computing power to allocate, and billing by invocation volume" into Stripe's familiar payment process: developers connect models through OpenRouter and complete deductions and settlements via Stripe, merging two originally separate infrastructures into a closed loop. In contrast, Coinbase connects USDC with AI agent payment scenarios through x402, aiming to occupy the entry into on-chain settlements so that every dollar payment initiated by AI goes through its protocol. One positions itself in the model distribution layer, while the other locates itself in the on-chain payment layer; the paths differ but the goals align: who can grasp the transaction entry in the AI era has the opportunity to define how money meets computing power in the next round of the gold rush.

Alphabet, AMD, and Intel Emerge as AI Shovel Sellers

Beyond the payment entry, capital is pricing another type of "shovel seller." Alphabet's early bet on Anthropic has led to its equity value reportedly soaring to about $124 billion, making it a significant asset within its portfolio. While Anthropic presents a narrative of AI safety and the Claude series models, what truly drives Alphabet's soaring book value is its control over essential elements like computing power, algorithms, and talent. The story's dividend ultimately settles into infrastructure assets rather than the short-term traffic of a single application.

Almost concurrently, the hardware end of shovel sellers is also actively amplifying its narrative. AMD announced the full production of its rack-level AI system Helios at the Advancing AI conference in San Francisco, directly targeting training and inference scenarios in data centers, while issuing a market expectation of approximately $1.4 trillion for the AI accelerator market by 2030, about $220 billion for data center CPUs, and around $2 trillion globally for computing, framing AI demand as a long-term computing power engineering order. Shortly thereafter, Intel released a better-than-expected Q3 revenue outlook (between $15.8 billion to $16.8 billion, higher than the market average expectation of $15.1 billion), causing its stock to surge nearly 13% after hours, interpreted as data center spending and AI demand beginning to bring true performance recovery to this traditional chip giant. In this round of the gold rush, it's still the companies responsible for "creating shovels" that are the first to realize rewards.

Tesla's Stock Plummets: Valuation Backlash from AI Narrative

In the same week when "shovel sellers" raised valuations based on real orders and performance, Tesla's second-quarter results served as a counterexample. Profits for the second quarter fell short of market expectations, and while specific breakdown reasons have not been publicly detailed, as soon as the financial report landed, the stock closed down over 14%, with S3 Partners estimating short-sellers made about $4.12 billion in profits that day. For many years, Tesla has reinforced its AI attributes in autonomous driving and humanoid robots, positioning itself as a "software company" and "robot company," but when actual profits do not match the narrative's height, valuation begins to backlash; the first to realize profits aren't the future industrial kingdom, but the capital betting against this narrative stock.

This plunge stands in stark contrast to the strengthening stock prices of chip and data center companies like AMD and Intel, buoyed by optimistic expectations driven by AI demand: one side has clear delivery rhythms, capacity planning, and long-term market scale guidance, and the market is willing to pay for "shovels"; on the other side, autonomous driving and robotics have yet to form stable, high-certainty cash flows, leading investors to question whether Tesla's AI narrative can convert into sustainable profitability within a reasonable timeframe. This round of severe correction has laid open the role differences within the AI ecosystem for everyone to see: those carving shovels are realizing results, while the storytellers are bearing the costs of valuation returning to reality.

The Next Game Between Crypto and Tech Under the AI Gold Rush

If we draw a line from the signals this week, we would see a very clear differentiation: on one end, Alphabet, AMD, and Intel solidify their "shovel seller" status with Anthropic's book gains, Helios production, and optimistic performance guidance; on the other end, Tesla's stock plunged significantly following disappointing profits, reminding the market that not all companies with AI tags can immediately deliver profitability. The crypto world is also undergoing simultaneous reshuffling: Odos enters the shutdown process, ringing the bell for the high-complexity, low-moat DeFi aggregation sector to exit; meanwhile, Coinbase connects USDC payments with AI agent scenarios through x402, and Stripe attempts to get closer to model distribution entry by acquiring OpenRouter, with new stories unfolding between on-chain payments and AI trading entries. Key observations moving forward include the progress speed of the on-chain payment and AI agent integration, and whether this path can transition from conceptual demonstrations to stable revenue; second, after projects like Odos exit, whether the DeFi sector will reorganize into a few protocols with genuine demand and clear charging models in a more thorough clearing; third, whether AMD and Intel can fulfill their optimistic expectations for the AI accelerator and data center markets by the 2030 timeline, and how Alphabet, through its substantial stake in Anthropic, will reshape its AI strategic landscape among foundational models, security, and application layers.

Join our community, let's discuss and grow stronger together!
Exclusive Hyperliquid benefits for AiCoin: https://app.hyperliquid.xyz/join/AICOIN88
Exclusive Aster benefits for AiCoin: https://www.asterdex.com/zh-CN/referral/9C50e2
On-chain Telegram community: https://t.me/AiCoinWhaleData
On-chain community: https://www.aicoin.com/link/chat?cid=N6OVMor5g
AiCoin on-chain Twitter: https://x.com/aicoinwhaledata

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。