Chipmakers are facing their sharpest pressure in months as traders reassess whether the first phase of the artificial intelligence boom has already delivered most of its gains to semiconductor suppliers. The pullback has hit memory and broader chip shares quickly, with DRAM-focused exchange-traded funds falling about 25% from their June 22 highs and a wider semiconductor ETF losing roughly 12% in two weeks.
The decline does not necessarily signal the end of the AI cycle. Instead, market action points to a shift in where traders expect the next round of profits to appear. Capital is moving away from companies that supply the hardware behind AI systems and toward firms that use that infrastructure to offer cloud services, software tools, automated platforms, and consumer-facing AI products.
That distinction matters for both equity and blockchain markets. The early AI trade rewarded companies tied to graphics processors, memory chips, data centers, and storage capacity. The next phase may reward businesses and digital networks that turn that infrastructure into revenue-generating applications.
For blockchain traders, the message is especially important. Tokens linked mainly to raw computing power, storage capacity, or server access could face growing pressure if the market concludes that the shortage of AI infrastructure has turned into an oversupply. Tokens tied to useful applications, data markets, automation, and financial services may be better placed if capital continues rotating toward software and usage.
Chip retreat points to a change in the AI trade
The semiconductor retreat has been broad. The Philadelphia Semiconductor Index has pulled back alongside memory-heavy products, while traders have sold shares of companies that benefited most from the first stage of AI infrastructure spending.
Micron Technology reported revenue of $41.4 billion and pointed to a forward target of $50 billion, numbers that would normally support a bullish view on memory demand. Yet its stock still declined. That reaction suggested traders had already priced in a strong earnings cycle and were beginning to look beyond current results.
The market’s response was not about weak numbers. It was about timing. When a company delivers record or near-record guidance and its shares fall, the signal is often that the market is no longer paying for what is happening now. It is asking what comes next.
In this case, the concern is that memory and chip suppliers may be approaching a cyclical peak after a period of unusually strong demand tied to AI data center construction. DRAM, high-bandwidth memory, GPUs, networking equipment, and storage systems all benefited as major technology companies raced to build infrastructure for large language models and AI platforms.
That spending surge created a powerful profit cycle for hardware suppliers. But it also raised the risk of overbuilding.
Meta’s computing plan sharpened the selloff
The warning became clearer after Meta announced plans to rent out surplus computing power. The move was widely read as a sign that one of the largest AI infrastructure buyers may have more capacity than it immediately needs.
For traders, that announcement mattered because cloud and social media companies have been among the biggest sources of demand for chips, memory, servers, and data center equipment. If one of those firms is prepared to lease out excess capacity, it raises questions about how much new hardware it will need in the near term.
The announcement did not mean AI demand has disappeared. It did suggest that the most aggressive stage of infrastructure buying may be slowing, at least temporarily. That was enough to trigger selling across chip stocks.
The effect was amplified by Samsung’s latest report, which showed that the sector’s profit surge was not limited to one company’s technology advantage. Rather, it reflected a broad industry upswing. That is positive when the cycle is rising, but it can become a concern when traders start thinking the cycle is mature.
If many companies benefited from the same demand wave, many could also suffer when that wave slows.
Profit expectations are moving down the AI chain
Morgan Stanley strategist Michael Wilson has argued that traders should reduce exposure to semiconductor stocks and favor large-scale cloud computing companies. His view is based on a transition in profit distribution across the AI supply chain.
In the first stage of the boom, profits flowed heavily to hardware suppliers such as Nvidia, Micron, and SK Hynix. These companies provided the components needed to build AI infrastructure. In the next stage, Wilson argues, earnings may shift toward companies that operate AI platforms, sell cloud services, and deliver software-based tools to customers.
That view fits recent market behavior. If the AI expansion were truly ending, traders would likely sell most AI-related shares at the same time. Instead, the pattern looks more like an internal rotation. Hardware makers are under pressure, while some downstream companies tied to applications and services are attracting renewed demand.
This rotation is common in major technology cycles. Early gains often concentrate in suppliers of scarce infrastructure. Later gains move toward companies that use that infrastructure to create products, services, and platforms with wider adoption.
The internet cycle followed a similar pattern. The first phase rewarded network equipment and fiber buildout. Later value moved toward search, e-commerce, cloud platforms, and software. AI may now be entering a comparable handoff.
Alibaba’s rally shows where some capital is heading
Alibaba’s recent 11% surge in U.S. trading stood out because it happened while semiconductor shares were weakening. The move suggested that traders were willing to buy companies positioned around AI services and cloud platforms, especially in markets where domestic technology ecosystems carry added strategic importance.
Alibaba is not simply a retail or e-commerce name in this context. Its cloud business, AI models, and enterprise services make it part of the broader AI application layer. The rally showed that capital may be looking for companies able to commercialize AI infrastructure rather than merely supply it.
The move also reflected geopolitical themes. As governments and companies seek more self-reliant AI systems, domestic platforms in major regions may gain strategic value. Reports of possible export controls on advanced AI models added to the reassessment of local technology leaders.
This does not mean every cloud or software company will rise immediately. U.S. cloud operators such as Microsoft, Google, and Amazon have not seen the same kind of short-term share movement as Alibaba. Their valuations are already elevated, and traders remain cautious about how slower capital spending could affect near-term profit margins.
Still, the broader market logic is clear. If spending on chips and servers cools, companies that already own large AI infrastructure may benefit from using it more efficiently. Over time, that could support cloud margins and software revenue.
Oversupply risk is now part of the AI story
The most important change in the market narrative is the move from scarcity to possible surplus.
During the first phase of AI enthusiasm, traders focused on shortages. There were not enough advanced chips, not enough high-bandwidth memory, not enough data center capacity, and not enough power infrastructure. Companies that could provide those resources enjoyed strong pricing power.
Now the question is whether too much capacity has been built too quickly.
Major social platforms and technology companies are expected to spend heavily on data centers, with some industry estimates pointing to spending of up to $145 billion this year among leading players. Such spending can support chip demand in the short run, but it can also create a later need to rent out unused capacity to recover costs.
Meta’s surplus computing plan fits that concern. If large companies begin offering excess AI capacity to others, the market price for computing power could fall. That would put pressure on businesses and networks whose main value proposition is access to raw compute.
This is where the signal becomes relevant for blockchain markets.
Blockchain traders face a different kind of AI risk
The drop in global technology shares carries a direct warning for blockchain traders. Tokens linked to decentralized physical infrastructure, server networks, storage markets, and compute access may be vulnerable if the underlying resource becomes less scarce.
Decentralized physical infrastructure networks, often called DePIN projects, attracted strong attention during the period when traders believed computing power, data storage, wireless coverage, and physical network capacity would remain in short supply. Sector trackers have reported that the total value of physical token networks crossed $32 billion nearly two years ago, showing how much capital moved into that theme.
The problem is that many of these networks depend on a scarcity story. If basic server space, storage capacity, or rented computing power becomes cheaper because large technology companies have overbuilt, the economic case for some tokens could weaken.
This does not mean every infrastructure-linked token is at risk. Projects with strong customer demand, clear pricing, reliable usage, and real revenue may still find a place. But tokens that mainly match buyers and sellers of generic hardware capacity could face pressure if centralized cloud providers flood the market with cheaper alternatives.
The same issue applies to storage tokens. If cloud companies and data center operators have too much capacity, decentralized storage networks must compete not only on ideology or token incentives, but also on price, reliability, compliance, and ease of use.
For traders, the key question is no longer whether a project owns or coordinates infrastructure. The question is whether that infrastructure is needed, used, and paid for at sustainable rates.
Application tokens may draw more attention
If capital continues moving away from raw infrastructure and toward usage, blockchain projects tied to applications may gain more attention.
That includes tokens linked to automated software programs, machine-to-machine payments, open data markets, identity tools, decentralized finance, and consumer services with clear daily activity. These areas are closer to the software layer of the AI economy, where platforms can build products on top of existing computing resources.
The same logic driving equity rotation applies here. Hardware and infrastructure benefit first when a new technology is being built. Applications benefit later when users arrive.
For blockchain markets, this could favor networks that solve practical problems rather than simply offer capacity. Projects that support payments, data licensing, AI agent coordination, tokenized assets, and user-facing financial tools may be more aligned with the next stage of AI adoption.
JPMorgan strategist Mislav Matejka has also warned that the current technology boom may not remain the only dominant trade. That view supports the idea that traders may need to broaden their focus beyond the first winners of the AI cycle.
In digital assets, that means looking beyond tokens tied to GPUs, chips, servers, or storage alone. The next phase may be more selective, with stronger performance concentrated in networks that show real usage and revenue potential.
The AI cycle is changing, not disappearing
The recent selloff in chip stocks should not be read as a simple rejection of AI. Demand for AI tools, automation, cloud services, and model deployment remains significant. Companies are still racing to integrate AI into search, advertising, software development, logistics, customer service, finance, and media.
What is changing is the market’s view of who captures the profit.
The first phase rewarded scarcity. The next phase may reward efficiency, distribution, and monetization. That shift can be uncomfortable because it turns former market leaders into sources of funding for new trades.
Semiconductor stocks may stabilize if earnings remain strong or if demand from cloud providers continues at a high level. But the recent price action shows that traders are no longer willing to assume that every AI hardware supplier will keep expanding margins forever.
The same caution applies to blockchain infrastructure tokens. If the world is moving from a shortage of machines to a surplus of machines, then the value of merely providing machine access becomes less certain.
What traders should watch next
The clearest signals will come from capital spending plans at major cloud and platform companies. If Microsoft, Google, Amazon, Meta, and other large operators slow data center expansion, chip and memory shares could remain under pressure. If they keep spending aggressively while also showing strong AI revenue, the hardware trade may recover.
Traders should also watch rental pricing for cloud compute, demand for AI model hosting, memory contract prices, and margins at major chipmakers. These indicators will show whether the market is facing a temporary pause or a deeper supply adjustment.
In blockchain markets, daily network usage, real fees, customer demand, and developer activity will matter more than broad AI branding. Tokens that rely only on the promise of future infrastructure demand may struggle if cheaper centralized capacity becomes widely available.
The broader message is straightforward. The AI trade is moving from construction to commercialization. That does not end the opportunity, but it changes the winners. Chipmakers powered the first leg of the boom. Cloud platforms, software companies, data markets, and useful blockchain applications may define the next one.
As capital rotates from chips to AI platforms, explore smarter automation with our AI copy trading tools for dynamic market positioning.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
