TSMC printed the most unambiguous AI capex endorsement of the year on Thursday, $40.2 billion in revenue, net income up 77%, CoWoS advanced packaging booked out more than a year, and a full-year AI growth outlook lifted above 40%. The Philadelphia Semiconductor Index responded by entering a bear market. If that dissonance bothers you, it should. But it is not a contradiction, it is the tape telling you something precise about where we are in the cycle, and what comes next.

"TSMC confirmed the AI infrastructure buildout is accelerating. The chip selloff is not about demand, it is about multiples, memory, and a market looking for its next leadership rotation."

The Foundry Said Yes. The Market Said "We Know." Now the Question Is Who Captures the Next Dollar.

Thursday's TSMC print was the fundamental verdict on the AI infrastructure thesis that the market has been debating all summer. Second-quarter revenue rose 36% year-over-year and 12% sequentially, while net income surged 77.4%, the ninth consecutive quarter of double-digit profit growth. Gross margin climbed to 67.7%, above the company's own guidance ceiling, reflecting the pricing power that comes from controlling roughly 73% of the global advanced foundry market. Most importantly, TSMC confirmed CoWoS advanced packaging is fully sold out, with lead times now exceeding one year. That is not a company in a cyclical plateau. That is a company whose capacity constraints are the binding throttle on the AI buildout.

And yet: semiconductor stocks officially entered a bear market this week, with the PHLX Semiconductor Index peaking at 14,655 in late June and sliding to approximately 11,674 by July 17, a decline of more than 20%. The damage is enormous: global chip stocks erased about $3.3 trillion in market value since June 22. The S&P 500 shed 1.55% on the week and the Nasdaq dropped 2.9%, while semiconductor shares fell approximately 10% in the week alone.

Here is Nova X's read, and it differs from the consensus: the selloff is not a demand story. It is a multiple story layered on top of two distinct catalysts, neither of which touches the AI capex thesis at its foundation. The market is repricing who captures the next dollar of AI spending, not whether that dollar gets spent.

Pillar 1 · AI Chips / Hardware (NVDA, AMD, AVGO): Bear Market in the Index, Not in the Thesis

Two catalysts drove the chip selloff this week, and both deserve a precise diagnosis. First, Chinese AI lab Moonshot launched Kimi K3, a 2.8-trillion-parameter open-weight model that its developers claim rivals top-tier models from OpenAI and Anthropic, reviving DeepSeek-era fears about AI commoditization. Second, a South Korean brokerage downgraded SK Hynix on fixed-price HBM contract concerns, sparking broad chip profit-taking that cascaded from memory into logic and equipment names.

Neither catalyst is about TSMC's backlog. Neither changes AVGO's hyperscaler custom silicon pipeline. What they do is give a crowded trade a reason to reduce. The semiconductor index remains over 60% higher in 2026, even after closing just above 20% below its June peak, and with substantial gains and leveraged trades in the complex, there is minimal tolerance for weaker guidance. That is not a fundamental break. That is a setup.

The non-consensus read: TSMC's earnings did not cause the chip selloff and were not priced by it. The shift in TSMC's revenue mix is the clearest single measure of how thoroughly AI has restructured global semiconductor demand, High-Performance Computing rose 20% sequentially in one quarter and now accounts for 66% of total wafer revenue, while smartphones fell to 22%. A 66% HPC mix is not a sector approaching saturation. It is a sector in mid-cycle acceleration. The market knows this. It is simply digesting three months of 105% gains at once.

Pillar 2 · Nuclear / Power Infrastructure (VST, CEG, LEU): Oil at $88 Is the Best Power Bull Thesis You Weren't Running

The Iran conflict escalated sharply this week, and the energy market's verdict was unambiguous. Oil prices surged more than 4% on Friday, with Brent crude settling at $88.10 a barrel and WTI closing at $82.49, both benchmarks posting weekly gains of roughly 16%. The fragile truce reached in June has collapsed, disrupting energy flows through the Strait of Hormuz, which handled about a fifth of daily global oil and LNG trade before the war began.

Iran is now deploying economic pressure on energy routes, asking Houthi forces in Yemen to prepare to shut down the Bab el-Mandeb Strait in the Red Sea. With the Strait of Hormuz already disrupted, this threat raises the serious risk of both of the Middle East's primary oil export routes being simultaneously impaired. The asymmetry matters: if both routes are disrupted at the same time, the global LNG market reprices overnight.

The power infrastructure thesis, that carbon-free, price-stable baseload energy commands a structural premium in an era of data center electricity demand, just got a second engine. The Iran conflict has bolstered the necessity for renewable and alternative energy, as distributed and nuclear power can reduce vulnerability to external supply disruptions, and with fluctuating oil prices, non-fossil baseload has become significantly more cost-competitive. VST and CEG are not oil hedges. But when Brent is at $88 and climbing toward a two-route supply disruption, the relative value of contracted, carbon-free power just increased materially. Watch for data center PPA acceleration on this backdrop.

Pillar 3 · Digital Finance Rails: The GENIUS Act Turns One, and the Rulemaking Clock Just Hit Zero

The most structurally significant event for the Digital Finance Rails pillar this week was not a price move, it was a deadline. The GENIUS Act attached a hard statutory deadline to its primary implementing regulations: July 18, 2026. That date landed today. President Trump signed the GENIUS Act into law exactly one year ago at a ceremony attended by members of Congress and leaders from major crypto companies; the law provides a framework for the regulation of US dollar stablecoin issuers at both the federal and state levels.

A year on, regulators are still working their way through the process of actually crafting the rules that stablecoin issuers will have to abide by. Markets have mostly priced the July 18 deadline as a legitimacy milestone for stablecoins, but industry observers read it as a cost-visibility event that decides which issuers can afford to keep operating. The consolidation dynamic is exactly what the institutional digital finance thesis requires: fewer, larger, better-capitalized issuers operating under bank-grade AML and reserve standards. GENIUS makes stablecoins safer to hold and easier for a bank or corporate treasury desk to justify, but that legitimacy carries a price, producing a market with fewer issuers, each one large enough to spread audits, licensing, and reserve management across billions in float.

The Clarity Act, which would extend the regulatory framework to the broader digital asset market, is now moving to the Senate. A House Financial Services subcommittee held a hearing Friday about what the Clarity Act will do for crypto, with the subcommittee chair stating that Congress needs to pass the bill to "replace regulation by enforcement with clear rules of the road for digital assets." The legislative flywheel is turning. The institutional on-ramp is being paved. The thesis remains intact; the fund is watching for a clean entry with the 30% bar that governs any rebuild in this pillar.

Pillar 4 · AI Software / Platforms (PLTR, META): The Rotation the Chip Bears Are Missing

Here is the underappreciated story inside the chip selloff: the capital is not leaving AI. It is leaving chip hardware for AI software and platform monetization. On Wednesday, investors appeared to pare exposure to key semiconductor stocks and move into shares of certain Big Tech names, Amazon and Alphabet moved up around 3%, Microsoft was higher by nearly 3%, and Apple gained 4%, hitting a new all-time high. That is not a rotation out of the AI trade. That is a rotation within it, from infrastructure hardware toward platform monetization.

Semiconductor positioning remains crowded despite the recent selloff, suggesting the correction may not be over if Microsoft, Amazon, Alphabet, or Meta begin emphasizing utilization, monetization, and capital discipline. The next phase of the AI trade may favor hyperscalers over chipmakers, with Alphabet's upcoming earnings offering the first major test of that rotation. For META, this dynamic is constructive: if Alphabet validates AI ad monetization on Tuesday, META gets a read-through into its own print later in the month. The AI ad monetization thesis is the most defensible near-term AI revenue story in the market right now.

Pillar 5 · Specialized GPU Cloud (CRWV): The Fragile Name in a Fragile Week

CRWV is the most leveraged name in the portfolio to chip sentiment, and chip sentiment just entered a bear market. The NVIDIA-backstopped contracted cloud model gives CRWV structural revenue visibility that pure-play hyperscalers lack, but in a risk-off week driven by AI commoditization fears, that distinction gets compressed in the tape. The thesis, contracted GPU access at scale for enterprise AI deployment, is intact, but the price action demands vigilance. The key question coming into next week: does the Alphabet capex guidance on Tuesday reaffirm that contracted GPU cloud demand is a feature, not a rounding error, in the hyperscaler build plan?

Alphabet Tuesday Is the Week's Only Number That Matters

The week ahead is the most consequential earnings gauntlet of the summer, and it opens with the single most important data point for every pillar in this fund. Alphabet and Tesla report on July 22. Alphabet is set to provide the most direct test of AI monetization when it announces results. Three questions govern everything else:

1. Capex guidance: Does Alphabet signal that hyperscaler AI infrastructure spending is on the $638 billion collective 2026 trajectory, or do they introduce the language of "capital discipline" that would detonate the chip complex further?
2. AI revenue: Is Google Cloud's AI revenue line growing fast enough to justify the capex? If yes, the AI monetization thesis (META, PLTR) gets a read-through. If not, the rotation-within-AI narrative breaks down.
3. CoWoS commentary: Any color on advanced packaging lead times from the hyperscaler buyer side validates TSMC's sold-out claim from Thursday's print.

The Fed is likely to keep interest rates on hold for the rest of 2026; the July 29 meeting will likely surface more from the FOMC on what data it is monitoring, but with inflation still contained and the labor market stable, the bar for rate hikes remains high. The rate environment is not the risk. The earnings environment is.

Over 80 S&P 500 firms are set to report results next week. So far, 90% of the initial 49 companies to report have surpassed expectations, a pace that would ordinarily be a tailwind. The problem is that the 10% that disappointed included IBM, which cratered 25% on soft software demand, and that the 90% beat rate has not prevented the SOX from entering a bear market. Beats against elevated expectations are table stakes. The market needs guidance acceleration to stop the chip multiple compression.

Conviction and What Would Change It

AI Chips / Hardware · Conviction: High. Watching the 200-day moving average.

TSMC's print is the thesis anchoring event of the quarter. The results surpassed analyst expectations across every key metric and confirmed that the infrastructure spending cycle many expected to plateau in 2026 has instead accelerated. The selloff is a crowded-trade unwinding driven by memory HBM contract fears and Moonshot AI commoditization anxiety, neither of which touches the CoWoS-constrained, HPC-dominant revenue profile that TSMC just reported. Ed Yardeni has warned that the S&P 500 Semiconductors index is likely to fall another 12% to its 200-day moving average, that level is the line we are watching. A close at the 200-day with intact capex guidance from Alphabet is the cleanest re-entry signal the tape can offer. What would change the thesis: an Alphabet earnings call that introduces "capital discipline" language paired with a reduction in forward cloud infrastructure commitments. That is not our base case.

Nuclear / Power Infrastructure · Conviction: High and rising.

Oil at $88 Brent, the Strait of Hormuz disrupted, and the Red Sea route under threat is not a temporary geopolitical blip, it is a structural repricing of energy security that makes carbon-free, domestically controlled baseload power more valuable in every conversation between hyperscalers and power developers. The data center electricity demand thesis was compelling when oil was at $70. At $88 with two shipping routes at risk, the urgency of long-duration PPA signings accelerates. VST and CEG sit at the intersection of that urgency. What would change the thesis: a diplomatic resolution that fully reopens the Strait and collapses oil below $75, removing the energy security premium from non-fossil baseload.

Digital Finance Rails · Conviction: Moderate, watching for the clean entry.

The GENIUS Act rulemaking deadline landing today is the most important structural event in this pillar since the law was signed a year ago. GENIUS turns stablecoin issuance from a crypto product into a regulated-scale business, and on July 18, issuers start finding out which side of that line they're on. The institutional on-ramp is being built, the Clarity Act is moving through the Senate, and the consolidation dynamic favors the large, well-capitalized players who can absorb bank-grade compliance costs. The fund is watching for a clean 30%-bar entry before rebuilding this pillar. The structural tailwind is intact; the timing is the only question.

AI Software / Platforms · Conviction: High. Alphabet Tuesday is the read-through.

The rotation from chip hardware into AI platform monetization that showed up on Wednesday's tape is the market's acknowledgment that the next dollar of AI return accrues to the software layer, not just the silicon layer. META's AI ad monetization thesis gets a direct read-through from Alphabet's Tuesday print. If Google Search and Cloud AI revenue growth accelerate, the case that AI is generating measurable platform revenue, not just capex, is confirmed. That is the thesis inflection point we have been waiting for since the spring.

Specialized GPU Cloud · Conviction: Moderate. Volatility is price, not thesis.

CRWV is the most volatile name in the fund's thematic universe and the most sentiment-exposed to the chip narrative. This week's chip bear market was the kind of tape that produces outsized price moves in contracted cloud infrastructure names with limited near-term earnings catalysts. The NVIDIA-backstopped revenue model remains the structural differentiator. What would change the thesis: a hyperscaler earnings call that signals migration of GPU workloads in-house, reducing demand for contracted third-party GPU cloud. That is not what the TSMC data is showing, CoWoS sold out for over a year implies external GPU cloud demand is, if anything, supply-constrained, not demand-impaired.

The week in one sentence: TSMC proved the AI buildout is real and accelerating; the chip index proved that a 105% rally in three months creates its own gravity regardless of fundamentals. Next week, Alphabet either gives the tape a reason to separate thesis from multiple, or it doesn't. That distinction is worth more than any single earnings line item.

This document is for informational purposes only and does not constitute investment advice, a solicitation, or an offer to buy or sell any security. The views expressed are those of Nova X Convergence Fund's research team as of July 18, 2026, and are subject to change without notice. Past performance is not indicative of future results. All investments involve risk, including the potential loss of principal. This letter is intended solely for the use of the recipient and may not be reproduced or distributed without written permission. Always consult a qualified financial advisor before making investment decisions.