This AI ETF Shows Why Your AI Portfolio Isn’t Making You Money

Photo of Omor Ibne Ehsan
By Omor Ibne Ehsan Published

Quick Read

  • Themes Generative Artificial Intelligence ETF (WISE) is down 4% year-to-date and up 14.6% over one year, significantly underperforming NVIDIA (NVDA), which is up 18% YTD and 62% over one year, because equal-weighting the index forces exposure to unprofitable software names like BigBear.AI and Gorilla Technology that dilute returns from mega-cap profit engines like NVIDIA and AMD (AMD).

     

  • The economics of AI have concentrated in hyperscaler infrastructure spending and capex beneficiaries like NVIDIA and AMD, rendering broad exposure across the full AI value chain via quasi-equal-weighting inefficient for investors with hardware conviction.

     

     

  • The analyst who called NVIDIA in 2010 just named his top 10 AI stocks. Get them here FREE.

Walk through a typical retail brokerage account in 2026 and the AI sleeve looks like a teenager’s bedroom. A little NVIDIA (NASDAQ:NVDA | NVDA Price Prediction), a little Palantir (NASDAQ:PLTR), a few “AI-adjacent” small caps somebody mentioned on a podcast, maybe a thematic fund bought because the ticker sounded clever. The Themes Generative Artificial Intelligence ETF (NASDAQ:WISE) is the cleanest case study of why that approach has quietly stopped working. WISE owns the right concept, companies whose revenue ties to generative AI, but the index blends mega-cap profit engines with money-losing startups, and the math of that blend is what an honest portfolio-fit conversation has to start with.

What WISE actually owns

WISE tracks the Solactive Generative Artificial Intelligence Index, launched in December 2023, holds 48 names, and charges 35 basis points. AUM sits near $31 million. Top weights include AMD (NASDAQ:AMD), QuickLogic (NASDAQ:QUIK), and Nvidia, with the top 10 making up 47% of the fund.

The index aims for broad exposure across the AI value chain, defensible in 2024 when nobody knew which layer would capture economics. Two years in, the layer that captured the economics is hyperscaler infrastructure, and a quasi-equal-weight index punishes you for that clarity.

It’s no longer worth your time to heavily buy AI startups because it’s obvious where things are heading. Not only are they bleeding cash, Wall Street isn’t going to irrationally dump money into them just because “they do AI”.

The return gap

WISE is down about 4% year-to-date and up about 14.6% over the trailing year. Nvidia over the same windows is up about 18% YTD and 62% over one year.

An investor who wanted AI exposure in 2025 and chose WISE got a fraction of NVDA’s returns. The fund delivers what its prospectus promises. It also loses to a generic tech stock.

Where hyperscaler capex actually lands

Nvidia reported Q4 FY2026 revenue of $68.13 billion, up about 73% year over year, and non-GAAP EPS of $1.62. FY2026 free cash flow ran $96.58 billion, and Q1 FY2027 guidance points to roughly $78 billion in revenue.

CEO Jensen Huang told investors “computing demand is growing exponentially, the agentic AI inflection point has arrived.”

Advanced Micro Devices posted Q1 FY2026 revenue of $10.25 billion (up about 38%), Data Center revenue of $5.78 billion (+57%), and a Meta agreement for up to 6 gigawatts of MI450 Instinct GPUs. AMD is up 79% YTD and 250% over the past year. WISE owns slivers of both. Its returns are dragged down by everything else in the basket.

The software side is where the dilution lives

In 2024 you could argue that small AI software names were cheap optionality on a build-out nobody fully understood. Now their financials are visible and most are not pretty. BigBear.AI and Gorilla Technology, both top-10 weights, are unprofitable and growing slowly enough that their multiples look like wishful thinking.

Microsoft (NASDAQ:MSFT), which actually rings the AI cash register (Azure +40%, a $37 billion AI run rate growing 123% year over year), is down about 10% year-to-date as the market repunishes capex-heavy AI software. If the cleanest software-side compounder is flat, speculative software names cannot rescue your AI sleeve.

The tradeoffs you are actually accepting

Three real costs of holding WISE.

  1. Equal-ish weighting mechanically caps your participation in the names doing the work. NVIDIA generated $96.58 billion in free cash flow last fiscal year. Owning it at low single-digit weight inside a basket of unprofitable peers dampens the trade you came for.
  2. The 0.35% expense ratio looks reasonable in isolation. You can get the largest WISE holdings for a few basis points inside QQQ, which has outperformed the thematic fund this cycle anyway.
  3. The roughly $31 million AUM is small enough to raise real survivorship risk. Thematic ETFs this size get closed when the theme cools, leaving holders with taxable distributions on whatever the fund is worth that week.

Who it fits, and who should look elsewhere

WISE makes sense for an investor who wants one line item labeled “AI” on a statement and accepts index-level lag in exchange for not picking names.

For anyone with conviction that the AI build-out is a hardware and hyperscaler story, which is what the cash flow statements have been telling you for six straight quarters, WISE dilutes the trade.

The cleaner approach in 2026 is direct ownership of the capex beneficiaries (NVIDIA, AMD, the networking and memory ecosystem) plus QQQ for diffusion.

Investors seeking small-cap AI exposure tend to find more verifiable revenue on the hardware side, and the sleeve typically stays small. The era of paying up for a software startup’s narrative deck is finished.

 

 

Photo of Omor Ibne Ehsan
About the Author Omor Ibne Ehsan →

Omor Ibne Ehsan is a writer at 24/7 Wall St. He is a self-taught investor with a focus on growth and cyclical stocks that have strong fundamentals, value, and long-term potential. He also has an interest in high-risk, high-reward investments such as cryptocurrencies and penny stocks.

Continue Reading

Top Gaining Stocks

MU Vol: 49,901,452
DXCM Vol: 6,420,139
KR Vol: 4,459,716
BDX Vol: 2,294,408
EQT
EQT Vol: 5,316,877

Top Losing Stocks

CTRA Vol: 73,319,495
EL Vol: 2,405,799
NEM Vol: 5,330,083
MRNA Vol: 3,322,584
ENPH Vol: 6,007,392