This episode starts with a follow-up on the fundamentals and then looks at the major indexes and hot stocks.
Nvidia's high-end GPUs are facing a historic supply shortage.
Blackwell systems are extremely tight.
Nvidia's Grace Blackwell systems, including GB300 and B300, are becoming harder and harder to obtain, and delivery times keep getting longer.
Widebush analysts said such severe supply tightness at this stage of the product life cycle is rare and reflects how hot the market is.
The supply-demand imbalance is driven by two things: rising demand for global AI infrastructure buildout, and bottlenecks in high-bandwidth memory and advanced packaging capacity for the Blackwell system.
Enterprise customers remain enthusiastic about AI deployment. Although the market once worried that events like DeepSeek would hurt investment enthusiasm, supply-chain checks show no enterprise AI deployment has slowed because of that.
Analysts put it bluntly: in fierce market competition, no customer wants to lose its place in line.
High-end chips are forming a two-track market.
At present, the market is clearly split.
The cutting-edge Blackwell chips are being allocated first to hyperscale customers like Microsoft and Google for frontier models.
Meanwhile, spurred by new export policy, demand for the previous-generation H260 chips has seen a second surge. With Blackwell leading and Rubin delayed, institutions expect the 2026 Blackwell platform to account for 71% of Nvidia's high-end GPU shipments, further consolidating its dominance. The next-gen Rubin architecture has already been sampled, but mass shipments are constrained by HBM4 certification and related challenges, so large-scale output will only gradually take over in the second half of the year.
AI infrastructure investment is still in the early stage. Widebush remains constructive on Nvidia and believes this global wave of AI infrastructure spending is still in its early development phase.
With continued investment from all sides, strong demand is expected to continue through 2026 and beyond, leaving ample room for future growth. Google and Nvidia are shifting some orders to Intel, and the global chip foundry landscape is changing as TSMC capacity becomes fully loaded and a new turning point emerges.
Driven by the AI boom, TSMC's leading-node wafers and advanced packaging lines are fully booked and unable to absorb huge order volumes. To reduce single-supplier dependence, top players such as Google and Nvidia are quietly engaging Intel as a backup foundry for high-end processors to address serious capacity gaps. Google has already placed a large foundry order first.
Among the alternative suppliers, Google is moving the fastest. After months of testing, it has already placed an order with Intel, asking it to produce more than 3 million in-house TPU chips in 2028. Forecasts suggest Google's total TPU output in the next two years will exceed 6 million chips, marking a key breakthrough for Intel's foundry business.
Nvidia is also evaluating next-gen GPU process technology.
Although Nvidia has not placed a formal order yet, it is actively testing Intel's technology to see whether it can be used in a new processor that integrates four graphics chips.
This R&D effort is tied to the 2028 Feynman architecture, and Nvidia has already started early tape-out tests on Intel's 18A process, which is comparable to 2nm, to verify mass-production feasibility.
Advanced packaging may become Intel's breakthrough weapon.
TSMC's packaging bottlenecks are becoming more obvious, while Intel's EMIB route is cheaper and is emerging as the key entry point.
Giants like SK Hynix are validating compatibility between their products and Intel's packaging process.
CHAA
Intel's pledge to expand capacity across multiple regions has further boosted customer confidence in it as an alternative, but rebuilding trust remains the biggest challenge.
Although the opportunity is rare, Intel still has to prove it can deliver high-volume, high-yield advanced processes. Its history of delayed mass production and losses makes major customers cautious. From early tests to actual core-order migration, Intel still faces huge hurdles if it wants to replicate the deep trust TSMC enjoys.
Diversified supply is reshaping the foundry landscape. The surge in AI compute demand has fully occupied TSMC capacity and forced tech giants to adopt diversified supply strategies. This not only gives Intel a rare chance at high-end foundry work, but also signals that the global chip manufacturing landscape is changing profoundly.
A more resilient, diversified supply-chain system may take shape in the future.
Amazon is spending billions on fiber, and Corning is becoming an AI infrastructure winner.
Amazon has locked in fiber supply with huge spending.
Amazon announced a multibillion-dollar agreement with Corning to buy fiber-optic cables and connectivity solutions, aimed at supporting its rapidly expanding data-center footprint across the U.S. and ensuring the underlying network connectivity required for AI infrastructure, further strengthening its cloud advantage.
The heavy investment will directly help Corning expand production and add 1,000 jobs at its North Carolina manufacturing plant. This is an expansion of Amazon's previous tens-of-billions cloud-infrastructure investment in the state and will also help local advanced manufacturing.
Corning stock has soared, and it is now called an AI winner.
As a key supplier of connectivity for AI infrastructure, Corning has strong momentum. Boosted by this news, its stock has risen sharply, with year-to-date gains already more than doubling. This century-old company is drawing strong demand from hyperscale data-center operators with its high-quality products.
Big tech firms keep tying up capacity with money, and in the AI compute race Corning has become a core supplier that all major tech firms want to secure.
After reaching a supply deal worth up to $6 billion with Meta earlier this year, it also received a $500 million investment from Nvidia last month to expand its fiber manufacturing partnership.
This underscores Corning's irreplaceable role in the AI supply chain.
Power-grid access is becoming a core infrastructure challenge.
As companies rush to build large data centers, they are facing multiple pressures.
The biggest challenge is power approval.
Because these energy-hungry facilities can strain the grid during peak hours, they must get approval from grid operators and utilities before they can connect smoothly.
They are also joining forces to cultivate technical talent.
To address the talent shortage in fiber manufacturing, Corning and Amazon plan to launch a joint training program.
The program is designed to train students in the skills needed for fiber manufacturing and related technical jobs, providing solid high-skill talent support for the long-term development of the U.S. AI supply chain.
Let's look at the market.
The Nasdaq rebounded today, at one point rising more than 2%, and ultimately closed up around 1.3%. The daily candle is still moderately strong and leaves a certain upper shadow, showing some disagreement between bulls and bears. The current area is also near the support zone we drew, so the rebound fits technical behavior.
On the five-minute chart, during the European session we asked everyone to watch for a breakout above a horizontal line. After the breakout, price moved to the resistance zone above, and after touching it, or near that line, it pulled back sharply. The market is now oscillating between these two horizontal lines. In the new session, watch whether the rally can continue or whether it falls again. A small bullish trendline can also be drawn to observe whether it remains valid.
This is the reality: every important support level gets a reaction.
Bulls naturally want a strong reaction, but a real bullish comeback needs strength, and it needs consecutive strength.
Otherwise it is just a drop, then a rebound, then another drop, or a rebound followed by sideways chop.
Can it keep falling? Of course.
Can it stop falling? Also yes.
So from a single candle, there is no clear bias.
Looking at the Dow, today's candle is a doji and nothing changed. Pre-market action was the same. On the five-minute chart, the European session was choppy, then price rebounded and touched the middle horizontal line before falling sharply, showing that bulls were not very committed.
The Russell's rebound was similar to the Nasdaq, a medium-strength green candle, but it also pulled back. If we take today's low as support, we should continue watching this range. It is much the same as the Nasdaq.
The S&P 500 also pulled back after an early spike, with the upper wick longer than the body. The selloff was pretty obvious. European pre-market action had rallied and then given it back. That is why the next news session is important: there is a slight tendency to stop falling, but that tendency is about balanced with the bearish one, so there is no clear direction yet.
Only the Nasdaq has shown a somewhat stronger rebound, and some tech leaders are bouncing meaningfully.
We will look at the leaders shortly. Gold is also range-bound here because the lower area is also a choppy support zone. For day traders, the move is basically a long setup in the U.S. session. Later we just watch whether this support zone remains effective. This area is resistance, that area is support. See whether it falls back again or moves higher, and whether, like the Nasdaq, it re-tests and then continues up or down. Because the market is not in a strong trend right now, up-and-down swings are normal.
Crude oil is messy too. News flow is all over the place, with talk of attack, then no attack, then more chaos. Oil gapped up today and pushed higher for a while, but then quickly pulled back. So this is only a day-trading market. Gap up, sharp rally, sharp pullback. The chart is not particularly meaningful.
Let's look at the overall U.S. market performance.
The leading sector today is clearly semiconductors. Software, internet, and consumer electronics are all down. Auto manufacturing, semiconductor equipment, and materials are also leading.
Computer hardware is okay too, which is basically our leading companies. But software and consumer hardware are not doing well.
Looking at the leaders, Applied Materials is one we've been tracking because it also sits in the optical/solar-related segment. Last Friday it was an inside candle; today it gapped up and ran immediately. So a chart that has not broken its trendline can still be fine, and you cannot assume the final down move is confirmed.
We discussed this over the weekend too: if a chart has not broken its long-term trendline, you must watch whether it continues to fall.
It did not fully break down, and on the five-minute chart it just ripped higher. That sort of move is annoying, right?
Next we keep watching whether the upside can continue. That is Applied Materials.
Now Micron. It looks similar. Its medium- and long-term trendline has not been broken, so we said to watch for a break on Monday or Tuesday. That did not happen. Then this morning, Jensen Huang said every dip is a good opportunity, which gave the market a lift. We also noticed pre-market that it did not truly break down, so there was no need to keep trimming aggressively.
On the five-minute chart, if you sold at that earlier level, you might buy back one or two sessions later. Why do it that way? Because you need to guard against further downside. Do not think it is impossible. It might extend lower at any step. That is why we act this way. Today it gapped up and then pulled back after hitting higher levels. If you buy back in the afternoon or near the close, that is also okay, but you still need to guard against another pullback.
On the daily chart, the long trendline has not been broken. Sandisk, however, did break its trendline. If you reduced exposure there, I do not think that was a problem, because it is still unclear what happens next. The S&P does not have the same strict directional line. Only tech stocks are showing this rebound tendency. On the five-minute chart, after the gap up, it clearly gave back gains. So if you are entering near the close, be aware that the market is still in a choppy zone. If it keeps rallying, it may simply be a high-level shakeout pattern. This is a hard chart to trade: you have to guard against risk and also avoid missing the move. A doji candle does not mean absolute safety.
Intel is another example. If it is on the right side of the downside circle, we may consider buying it back. It really formed an ABC pattern, meaning the double top failed. A double top can fail or succeed. If you sold earlier, you should have bought back here later. Sometimes that feels like a wasted trade, but it is not. That is simply how hard this is. People who cling to long positions may accidentally gain a lucky break, but that is not a given; charts can also keep falling.
These moves are all about guarding against further downside. They are not random actions. Eventually you will see that this kind of process has huge value. That is it.
Then there are Tesla-related names. Let's go back to our hot list. Many charts are falling. Tesla, for example, also closed higher today, but today's green candle was not as large as Friday's red candle, suggesting that more sideways chop is still possible. On the five-minute chart we are watching the possibility of an ABC decline, but there is still a downside trendline waiting for confirmation. So be patient and see what happens next.
Broadcom's rebound was not ideal either, only 2.8%. Today it simply gapped up and then faded. So do bulls really want to come back, or not? There is clearly some disagreement. If we look at the five-minute chart, you would try to go long if you strictly follow the right side of the downside pattern, but the stop-loss still belongs below. The bullish force today is not that strong. The chart just gapped up and then chopped sideways all day. If you try again, that is fine, but keep the stop below. If price keeps rolling over, you need to step out again.
The daily chart is the same. TSMC is the same too: today's move was not a very strong green candle. The five-minute chart is also similar. You can try going long, but the stop still needs to stay below. There is still a chance of further downside. Like Micron, it has not broken our trendline or the rising trendlines beneath it. If it does break, it could turn into a wide-range consolidation.
Nvidia is also the same: it gapped up and then pulled back. That pullback is really a sign of disagreement, meaning some traders are still cautious. On the five-minute chart, it did not keep pushing up to break the horizontal resistance line. Instead, it was blocked by that resistance. So bulls are not completely safe yet. On the daily chart, we see a red candle, but it was mainly caused by the gap-up open. There are still sellers trimming positions.
Amazon closed red, and many other charts also closed red. Apple had a big spike and then a big fade. It is still hard to trade this market. These charts spike and then drop sharply. Annoying, right?
So now the market is still in a state of sideways chop or pullback, and a medium pullback or even a deeper pullback is still possible. Everyone should still be alert to charts like Micron: can they hold at the top?
