Thematic Pulse: Investors Are Buying the AI Selloff—But Not the Entire Innovation Complex

Thematic Pulse: Investors Are Buying the AI Selloff—But Not the Entire Innovation Complex

Nvidia’s earnings provide fresh support for AI infrastructure demand, but ETF flows reveal a more selective investment landscape. Investors are accumulating semiconductor and disruptive-technology exposure while abandoning software, speculative biotechnology, and several capital-intensive themes. Meanwhile, some of the strongest recent performers are experiencing redemptions.

Thematic Investors Are Buying Weakness, but Conviction Is Concentrated

The thematic investment complex enters Wednesday with a notable disconnect between price performance and investor behavior. Across 155 ETFs with available returns in our selected thematic universe, 126 have declined over the trailing month, while 154 registered losses over the past week. The median one-month decline is approximately 5.0%, compared with losses of 7.5% for QQQ and 4.0% for SPY.

Yet flows suggest investors are not uniformly retreating from innovation. They are concentrating capital in a relatively small number of exposures where falling prices have created opportunities to reestablish positions in long-term growth themes.

Four ETFs—SMH, SOXX, ARKK, and ARKW—attracted approximately $14.0 billion in combined one-month inflows despite losses ranging from 12.8% to 14.8%. Excluding those four funds, the remaining ETFs in our selected thematic universe experienced approximately $6.7 billion in net redemptions among funds reporting flows.

That concentration matters. Investors appear willing to buy the AI correction, but their enthusiasm has not broadened into a general recovery in speculative growth assets.

Thematic Performance vs. Fund Flows

Trailing returns and net flows as reported in the November 18 data package; market performance through November 17 close.

ETF / Theme 1M Return 1M Net Flows 1W Net Flows
SMH – Semiconductors -14.8% +$5.11B +$108M
SOXX – Semiconductors -14.0% +$2.79B -$4.68B
ARKK – Disruptive Innovation -13.2% +$5.08B +$6.17B
ARKW – Next Generation Technology -12.8% +$1.02B +$1.12B
BAI – Artificial Intelligence -17.1% +$663M +$441M
KWEB – China Internet -9.6% +$437M +$2M
BUG – Cybersecurity -2.5% +$101M +$82M
IGV – Software -8.2% -$1.48B -$309M
ARKG – Genomics -13.3% -$1.72B +$127M
XBI – Biotechnology +0.7% -$249M -$138M
COPX – Copper Miners -18.3% -$310M -$47M
NLR – Nuclear Energy -10.6% -$133M -$51M
TAN – Solar -8.9% +$178M -$10M
ITB – Homebuilders +6.2% -$720M -$43M
XHB – Homebuilders +4.4% -$167M -$10M
PHO – Water Infrastructure +7.8% -$2M -$4M

Source: ETFThemes.com ETF returns and fund flows, November 18, 2026. Flows are rounded. Missing observations are excluded from aggregate calculations, and duplicate ETF tickers are counted once. One-month returns and flows cover matching trailing periods.

Nvidia Reestablishes the AI Demand Case, but Financing Is Becoming More Selective

Nvidia’s November 17 earnings report delivered important support for the fundamental AI investment thesis. Third-quarter revenue reached $115.8 billion, with data-center revenue rising 110% from a year earlier. The company projected fourth-quarter revenue of approximately $130 billion, while management highlighted strong demand and the rapid expansion of its Vera Rubin platform.

The results challenge the proposition that the recent decline in AI-related equities reflects an imminent collapse in end-market demand. They also explain why investors have been accumulating semiconductor ETFs during a significant correction.

However, the more consequential development may be Nvidia’s increasing selectivity in supporting the companies financing AI infrastructure.

The company’s willingness to provide investment commitments and financial support to neocloud operators has helped expand the AI ecosystem. Management now appears more focused on financing quality and the conditions under which those commitments become effective. Separately, reports that major banks are strengthening export-control requirements in data-center lending agreements suggest financing standards are becoming more discriminating.

This introduces an important distinction between AI demand and the profitability of financing that demand.

Semiconductor suppliers with strong customer commitments and established cash-generating businesses may remain attractive even if smaller infrastructure operators encounter more expensive financing. Conversely, companies dependent on leverage, repeated equity issuance, or Nvidia-backed financing could face a widening cost-of-capital disadvantage.

The ETF data reflect some of this uncertainty. SMH recorded $5.11 billion of one-month inflows, but SOXX experienced $4.68 billion of redemptions over the latest week despite positive one-month flows. Investors have not established uniform conviction even within semiconductors.

Importantly, Nvidia’s earnings were released after Tuesday’s market close, so the November 18 ETF dataset does not capture the subsequent market response.

Cybersecurity and AI Applications: Similar Growth Narratives, Different Investment Signals

The week’s technology disruptions across enterprise software, aviation, banking, and government systems have reinforced the economic importance of operational resilience. Although officials have not identified evidence of a coordinated cyberattack, the incidents highlight the complexity of maintaining and securing increasingly interconnected technology infrastructure.

That backdrop is potentially constructive for cybersecurity, systems monitoring, identity protection, and software infrastructure management.

Investors are already differentiating these exposures. BUG attracted approximately $101 million over the past month despite a 2.5% decline, while HACK received roughly $73 million with a 3.8% loss. CIBR experienced one-month redemptions but drew approximately $143 million over the latest week.

Traditional software has received a much less favorable response. IGV lost 8.2% over the past month and experienced nearly $1.5 billion in redemptions.

The distinction is fundamental. Cybersecurity addresses a continuing operating requirement, whereas portions of the application-software industry are confronting uncertainty about how autonomous AI agents will alter licensing models, customer relationships, and competitive advantages.

Recent agentic developments from Microsoft and Google strengthen the long-term enterprise AI adoption case. They may also undermine existing application providers whose products become features within larger AI platforms.

For portfolio managers, the opportunity lies in distinguishing companies that capture incremental AI spending from businesses whose existing revenues could be displaced by AI adoption.

Where Investors Are Selling Strength

Homebuilding provides the clearest example of investors using relative strength to reduce exposure.

ITB gained 6.2% over the trailing month but experienced $720 million in redemptions. XHB advanced 4.4% while losing another $167 million. These gains are notable against the broader weakness in growth equities, but the flows suggest investors are reluctant to extrapolate the recovery.

The macroeconomic explanation is straightforward. The Federal Reserve maintained its 3.75%–4.00% target range in October, with two policymakers favoring a rate increase. The prospect of additional tightening, reinforced by firm inflation data, complicates the outlook for mortgage affordability and housing activity.

Water infrastructure is another area where price performance has been stronger than investor enthusiasm. PHO gained 7.8% and FIW advanced 8.1%, but both experienced modest net redemptions. The withdrawals are small compared with homebuilders, suggesting profit-taking rather than a major abandonment of the theme.

Biotechnology reveals an additional form of selectivity. IBB gained 2.5% with approximately $191 million in inflows, while XBI gained just 0.7% and experienced $249 million in redemptions. More speculative genomics exposure was considerably weaker, with ARKG declining 13.3% and losing $1.72 billion over the trailing month.

Investors appear more comfortable owning established biotechnology businesses than assuming financing and commercialization risk in early-stage innovation.

Which Themes Benefit From the Latest Developments?

AI infrastructure and semiconductor manufacturing remain fundamentally supported, particularly where earnings visibility is grounded in confirmed orders and financially strong customers. Nvidia’s latest results support this conclusion, but the renewed attention to financing conditions argues against indiscriminate exposure across the AI supply chain.

Cybersecurity and operational resilience are becoming relatively more attractive. Recent disruptions should encourage spending on reliability, cloud security, monitoring, and recovery capabilities. The strength of the investment case depends on whether additional spending translates into recurring revenue and sustainable margins rather than temporary project activity.

Enterprise AI agents represent an expanding opportunity, with important competitive risks. Microsoft’s and Google’s latest developments reinforce demand for automation, inference, and enterprise AI deployment. Infrastructure and software providers able to monetize these capabilities could benefit, while legacy applications vulnerable to functional substitution may face pressure.

Selected biotechnology franchises are also comparatively resilient. Investors are favoring larger, better-capitalized businesses over speculative platforms, a distinction consistent with the broader preference for demonstrable earnings and manageable financing needs.

Which Themes Face the Strongest Headwinds?

Electrification, nuclear energy, and clean power remain exposed to the tension between favorable long-term electricity demand and less favorable near-term financing conditions.

NLR declined 10.6% with $133 million in redemptions, while GRID fell 8.7% and lost approximately $139 million. Solar presents a contrasting signal: TAN attracted $178 million over the month despite an 8.9% decline, although the latest week showed modest outflows.

These markets still benefit from structural demand associated with electrification and data-center expansion. However, higher capital costs, project delays, and uncertainty around development economics are weighing on investor confidence. Long-term demand growth does not necessarily translate into attractive near-term returns for equity holders.

Copper and industrial materials are experiencing a more decisive rejection. COPX declined 18.3% with $310 million in redemptions, reflecting concern about cyclical demand and the timing of infrastructure-related investment. Unlike semiconductors, copper mining has not attracted significant contrarian accumulation in the available data.

China-linked technology remains a policy-sensitive recovery trade. KWEB attracted approximately $437 million while losing 9.6%. Tencent’s improving AI-cloud demand offers a favorable fundamental development, but renewed export restrictions and uncertainty surrounding US-China negotiations complicate the outlook. Strong domestic AI adoption does not necessarily translate into improved earnings for companies whose access to advanced hardware is constrained.

Energy faces a different challenge. Despite renewed US-Iran hostilities, markets remain focused on improving oil supply conditions. XOP lost 11.4% over the month while attracting $148 million, although the most recent week showed redemptions. That looks more like opportunistic buying after a sharp decline than confirmation of a durable improvement in oil fundamentals.

Defense ETFs have also failed to attract substantial new capital despite geopolitical tensions. ITA, PPA, and XAR collectively experienced approximately $944 million of one-month redemptions. Geopolitical risk alone is proving insufficient to reverse profit-taking and broader portfolio repositioning.

Finally, blockchain and cryptocurrency-related equities remain among the weakest areas of thematic investing. BKCH fell 26.0% over the month and DAPP lost 20.5%, accompanied by redemptions. These exposures remain particularly vulnerable to tightening liquidity and diminishing demand for speculative risk.

Investment Outlook: Accumulation Is Not Yet Confirmation

This week’s thematic investment picture is becoming more differentiated.

The most compelling opportunities are emerging where recent price declines coincide with improving business fundamentals, measurable demand, and sustainable financing. Nvidia’s earnings support portions of the AI infrastructure complex, while technology resilience and enterprise automation provide additional avenues for growth.

However, ETF flows alone cannot establish that an investment has become fundamentally inexpensive. A 15% price decline is attractive only if forward earnings expectations, margins, and the cost of capital justify the resulting valuation. The current dataset establishes price and positioning changes, not a comprehensive forward-earnings valuation discount.

The upcoming FOMC minutes and subsequent inflation and labor-market developments will be important for determining whether investors can extend their buying beyond the most established AI exposures.

For now, we favor selective accumulation in financially sound AI infrastructure, mission-critical cybersecurity, and enterprise automation, while remaining more cautious on highly leveraged AI infrastructure operators, speculative biotechnology, China-sensitive hardware, and capital-intensive energy-transition exposures.

The critical question is no longer whether investors are prepared to buy the thematic correction. Clearly, some are.  The question is whether that capital can move beyond a handful of heavily accumulated ETFs and support a broader, earnings-driven recovery across the thematic investment universe.  The Fed’s stance on tightening from today’s FOMC minutes is the first piece of the puzzle

Sources

 

Disclaimer:  This commentary is provided for informational and educational purposes only and does not constitute individualized investment advice or a recommendation to buy or sell any security. ETF performance and flow data are historical and may change. Investment risks include market volatility, valuation changes, liquidity conditions, and regulatory developments.

Scroll to Top