Nasdaq hits record highs, AI funds abandon hardware, software rallies

Nasdaq hits record highs, AI funds abandon hardware, software rallies

Nasdaq still at records, S&P slowly rising, but the real issue is not whether it's up, but who's up. The market is no longer a broad rally but a highly concentrated, structured AI-driven market. Funds are draining liquidity from the entire market and pouring into a few core assets at unprecedented speed. More importantly, this change often doesn't appear in crashes but in the 'safe moments' of new index highs. In other words, when you think everything is fine, the structure has quietly changed. First, who is really driving the Nasdaq now? Is it a broad rally or just a few heavyweights controlling the index? Second, what level of capital migration is happening in the AI supply chain, and how do software and chips interact? Third, and most importantly, why might this AI rally be shifting from a 'liquidity-driven bull' to a 'structural differentiation bull'? If you're still thinking with the old 'buy the index and be safe' mindset, you may already be at a fork in this market shift. Next, we dive into data to thoroughly dissect this structural change. At the US close on May 28, the Nasdaq Composite closed at 26,917.47, up 0.91% (up 242 points), marking the 17th closing record high of 2026 and the 6th consecutive positive day. The S&P 500 closed at 7,563.63, up 0.58%, also six consecutive positive days. But the Dow Jones only rose 0.05%, barely participating in this core rally. This set of data together clearly indicates one thing: the market is not broadly rising but being forcibly pulled by a few mainline assets. And that mainline has only one answer: AI. The question is, how extreme is this AI rally? Let's look at the first real trigger today. Snowflake surged 36% in a single day, its largest single-day gain ever. But if you think it's just earnings beating expectations, you're seeing too shallow. Not only did earnings and guidance beat across the board, but it also signed a five-year AWS procurement agreement totaling $6 billion. The essence of this move is not just growth but locking in long-term demand for AI infrastructure. The market reacted directly. Software ETF rose 3%. Microsoft rose 3%, ServiceNow rose 6%. Chip side fully linked: Qualcomm up 6%, AMD up 5%, Sandisk up 6%. You'll notice a critical change: AI is no longer a single-point play but a chain reaction system. Software up drives cloud; cloud up drives servers; servers up in turn drive chips. The whole chain is starting simultaneously. But what really changed market expectations is not Snowflake but the second deeper change. Anthropic is redefining the entire AI industry's power structure. Latest market data is critical: Anthropic's annualized revenue has reached about $47 billion, at least 35% higher than OpenAI. More importantly, growth rate: Over the past five months, Anthropic's revenue grew 5x, while OpenAI only about 50%. This is no longer who leads but has entered a stratification stage. Look at profitability: Anthropic has achieved about 5% operating margin, while OpenAI still suffers over 100% loss structure. The core meaning is clear: AI industry has seen its first clear watershed. On one side, AI companies starting to make money; on the other, those still relying on financing for expansion. And Wall Street is most sensitive not to technology but to cash flow. So the market is doing something crucial: repricing the entire AI supply chain. If the core logic of AI over the past three years was technology-driven, now it's rapidly shifting to cash-flow-driven. Whoever can make money has pricing power. Whoever can earn stable profits gets valuation premium. That's why recent AI asset volatility structure shows clear divergence: some stocks are rising on valuation, some on earnings, and capital only rewards the latter. Third key change comes from the hardware side. Dell AI servers have entered explosive growth. AI server revenue reached $16.1 billion, up 757% year-on-year. AI orders reached $24.4 billion. The company even raised its full-year guidance to $60 billion in AI server revenue, with total revenue midpoint of $167 billion. After hours, it rose nearly 40%. This data clearly shows AI is not cooling but entering a second phase. The first phase: model explosion. The market competed on whose model is stronger, bigger parameters, who breaks capability boundaries first. OpenAI, Anthropic, Google, Meta all competed in the same logic: 'make AI smarter'. Capital flooded in, telling the capability upgrade story. But the problem: when model capabilities start to homogenize, the market naturally moves to the next step. The second phase: infrastructure realization. The logic is completely different: no longer 'who is smarter' but 'who can carry this intelligence'. Servers, storage, networking, data centers, power systems, and most critically GPU computing power shift from concept to real orders, from expectation to delivery. In other words, the first phase was building brains. The second phase is building the entire nervous system and energy system that the brain relies on. Once this phase starts, AI is no longer just a tech story but becomes a true infrastructure cycle. Capital shifts from storytelling to competing on capacity, delivery, and real cash flow. Servers, storage, networking, chips are now simultaneously entering volume cycles. But problems arise. When infrastructure fully explodes, what will the market do next? Answer must be divergence. Capital will swing between two directions: continue betting on hardware expansion, or start betting on software efficiency. Today's market already gave an important signal. Because the fourth variable comes from the macro level. US-Iran reached a temporary ceasefire agreement, extended 60 days. Risk appetite rebounded instantly. US Treasury yields fell, oil prices dropped, growth stocks quickly recovered. Plus inflation data itself. April PCE was 3.8%, core PCE 3.1%, exactly in line with expectations. Consumer spending grew 0.4%, employment remained resilient. The core meaning of this set is simple: inflation not out of control, growth not collapsed. This is the environment US stocks love best: easing rate pressure, declining risk, growth assets regaining pricing room. Putting today's four core signals together, you see a very clear market structure. Snowflake's 36% single-day surge directly ignited the software side, AI rally no longer limited to a few leaders but spreading along software, cloud, chips, servers, whole AI chain showing synchronized linkage. Anthropic's data started repricing the entire AI industry, its lead in revenue growth and profitability making the market distinguish for the first time between 'AI that makes money' and 'AI that still burns money'. Industry formally enters profit differentiation stage, valuation logic shifts from technology narrative to cash flow. Dell's AI server business exploded, with orders and guidance significantly raised, indicating AI infrastructure has entered real delivery cycle. Servers, storage, networking, computing power are entering volume phases together, hardware cycle accelerating. Macro environment simultaneously improved: US-Iran ceasefire eased geopolitical risks, inflation data in line, reducing concerns about rising rates, risk appetite returning, growth stocks and AI assets seeing capital inflows. When these four factors appear together, a very clear structure forms: AI is no longer just a single mainline but is repricing the entire Nasdaq. And the combination of these four points points to one common result: Nasdaq is not simply rising but being reshaped by AI. But the truly thought-provoking question is just beginning. When AI becomes more efficient, when software boosts computing power utilization, when one GPU can complete three times or more work than before, for Nvidia in the future, will demand continue to explode or be gradually diluted by efficiency? When Anthropic enters profitability cycle while OpenAI still runs at high loss, will the market reorder the valuation system of the entire AI ecosystem? When infrastructure investment peaks, will the next round of capital flow into completely different directions? This is the core of Wall Street's real debate now. Finally, I'll leave you with one question: If in the next year, AI truly enters a phase where efficiency replaces computing power, how long do you think this GPU-driven bull can last? Or will the next real winners shift from hardware to software? We'll keep tracking the real capital flows on Wall Street, not market surface noise. Let's dissect an even more critical question: Where is AI capital pulling out from and where is it re-concentrating? Note: If you're still using your AI understanding from three years ago to view today's US stocks, you may have completely missed the real core change of this rally. The problem is no longer how much Nasdaq has risen or whether S&P is hitting new highs, but a more critical structural change is happening: the market is completely switching from a 'liquidity-driven broad bull' to a 'structural bull driven by AI efficiency and cash flow'. Once this switch is established, the era of blindly buying leaders to make money may be entering its final stage. Have you noticed that indices are still rising, but money is no longer evenly distributed? Capital is converging on a few AI-related nodes at unprecedented speed, and it's moving together from software to infrastructure to chips. This is the most dangerous and most easily overlooked aspect today. If you only look at the indices, you'd think it's just another 'steady new highs' ordinary trading day, but the real issue is that the current rise is not a broadly diffused bull but a highly concentrated structural rally in a few core AI assets. Let's look at May 28 US close: Nasdaq Composite 26,917.47, +0.91% (+242 points), 17th closing record of 2026, 6th consecutive positive day. S&P 500 7,563.63, +0.58%, also six consecutive positive days. But notably, Dow Jones barely participated, only +0.05%. Together, this set conveys only one signal: the market is not broadly rising but being forcibly pulled by a few core assets, and that core force is only AI. In other words, US stocks have entered a very typical structural market: indices look strong, but internally highly divided. Money is not flowing in evenly but extremely concentrated into the AI supply chain. From software to cloud to servers to chips, the entire chain is being activated simultaneously. The true key to this rally is not how much it rose but that the structure changed. Because once the structure changes, all subsequent trading logic changes. But Dow Jones only rose 0.05%, barely participating in this core rally. If you break down this set, there's only one conclusion: the market is no longer a broad bull but a few heavyweight assets pulling the entire index, and the core driving force is only AI. If you still think AI is just a thematic investment, you must re-understand the changes ahead. Let's start with today's most direct trigger. Snowflake surged 36% in a single day, its largest single-day gain ever. But what really matters is not the gain itself but the structural change behind it. This company not only beat earnings and guidance across the board but also signed a five-year AWS procurement agreement totaling $6 billion. Note: this information is critical. It's essentially not a software company growth story but a signal of long-term demand locking for AI infrastructure. The market reacted directly. Software ETF up 3%. Microsoft up 3%, ServiceNow up 6%. Chip side also driven: Qualcomm up 6%, AMD up 5%, Sandisk up 6%. You'll see an important change. AI is no longer a single-point explosion but a complete chain reaction system. Software up drives cloud; cloud up drives servers; servers up drive chips; chips in turn strengthen the entire ecosystem. But what really changed market expectations is not Snowflake but the second deeper variable. Anthropic is redefining the AI industry's power structure. Latest market data is critical: Anthropic's annualized revenue has reached about $47 billion, at least 35% higher than OpenAI. More importantly, growth rate: over the past five months, Anthropic revenue grew 5x, OpenAI only about 50%. This is no longer leading but clearly entering stratification. Look at profitability: Anthropic has achieved about 5% operating margin, while OpenAI still suffers over 100% loss structure. The meaning here is clear: AI industry has seen its first watershed. One side: AI companies approaching real profitability; the other side: still relying on continuous financing and high-intensity expansion. And Wall Street is most sensitive not to technology leadership but to cash flow structure. So the market is doing something very important: repricing the entire AI supply chain. If the core logic of AI over the past three years was technology-driven, now it's rapidly shifting to cash-flow-driven. Whoever can make money has pricing power. Whoever can earn stable profits gets valuation premium. Third key change from hardware: Dell AI servers explosive growth. AI server revenue $16.1 billion, up 757% YoY. AI orders $24.4 billion. Company raised full-year guidance to $60 billion AI server revenue, total revenue midpoint $167 billion. After hours up nearly 40%. This data clearly shows AI is not cooling but entering second phase. Phase one: model explosion, relying on concepts and expectations. Phase two: infrastructure realization, relying on real orders and capacity release. Servers, storage, networking, chips – these links easily overlooked after model frenzy are now entering volume cycles simultaneously. Key point: this round is not single-point expansion but the entire AI infrastructure chain accelerating in sync. This means capital is truly shifting from storytelling to buying capacity. But then problems appear. When infrastructure fully explodes, the market will inevitably diverge. Capital will reallocate between two directions: continue betting on hardware expansion, or start betting on software efficiency. Today's market already gave a preliminary answer. Because the fourth variable: macro level. US-Iran temporary ceasefire agreement extended 60 days. Risk appetite instantly rebounded. US Treasury yields fell, oil prices dropped, growth stocks quickly recovered. Plus inflation data: April PCE 3.8%, core PCE 3.1%, exactly in line. Consumer spending +0.4%, employment resilient. Core meaning in one sentence: inflation not out of control, growth not collapsed. This is US stocks' favorite environment: easing rate pressure, falling risk, growth assets regaining pricing room. Putting today's four core signals together, you see a very clear structure. First, Snowflake ignited software side, AI chain starts spreading across the market. Second, Anthropic repricing AI leaders, industry formally entering profit differentiation. Third, Dell pushes AI infrastructure into new explosion cycle, server and computing power demand amplify simultaneously. Fourth, macro environment marginally improves, risk appetite returns, growth assets regain capital inflows. These four factors together point to one result: Nasdaq is not simply rising but being reshaped by AI. But the truly thought-provoking question is just beginning. When AI becomes more efficient, when software continuously improves computing power utilization, when one GPU can complete three times or more work than before, for Nvidia in the future, will demand continue to explode or be gradually diluted by efficiency? When Anthropic enters profitability cycle while OpenAI still runs at high loss, will the market reorder the valuation system of the entire AI ecosystem? When infrastructure investment peaks, will the next round of capital flow into completely different directions? This is the core of Wall Street's real debate now.