Vertigo3d
I came across the Invesco AI and Next Gen Software ETF (NYSEARCA:IGPT) entirely by accident. As a macro analyst, I keep track of different investment themes via the monitoring of ETFs. One particularly ETF that I regularly look at is IGPT’s predecessor, the Invesco Dynamic Software ETF (“PSJ”).
PSJ was a broadly diversified ETF that was not dominated by Microsoft (MSFT), the 800 lb gorilla in software, so it was a useful gauge of sentiment towards the industry (Figure 1).
Figure 1 – PSJ sub-industry allocation (invesco.com)
However, recently, I noticed that the PSJ was renamed into the Invesco AI and Next Gen Software ETF. This piqued my curiosity, since PSJ was a well established ETF dating back to 2005.
Is the newly renamed IGPT ETF a good way to track the ‘Artificial Intelligence’ theme?
Fund Overview
The Invesco AI and Next Gen Software ETF, as its name propose, aims to give investors exposure to “companies with significant exposure to technologies or products that contribute to future software development” themes admire artificial intelligence (“AI”) and robotics. Even the fund’s ticker, IGPT, is a play on the highly popular Chat-GPT app from OpenAI.
IGPT tracks the STOXX World AC NexGen Software Development Index (“NexGen Software Index”). Unlike rules-based market cap or sector indices, the NexGen Software Index selects stocks that the index provider believes are associated with the AI software and robotics theme.
Looking through the index construction methodology, the index provider has determined that companies within the subsectors shown in Figure 2 are eligible for the NxGen Software Index.
Figure 2 – Subsectors included in NexGen Software Index (invesco.com)
Stocks within the selected subsectors are ranked in descending order by free-float market cap, and the top 100 companies are selected for the index. Individual constituent weights are capped at 8% and the aggregate weight of stocks with weights greater than 4.5% may not exceed 45%. Any excess weight is redistributed to the remaining constituents of the index.
The index composition is reviewed annually in June and the index is rebalanced quarterly.
Portfolio Holdings
Figure 3 shows the top 10 holdings of the IGPT ETF. The IGPT is very concentrated, with the top 10 holdings accounting for 61.0% of the fund.
Figure 3 – IGPT top 10 holdings (invesco.com)
It includes the usual suspects that are associated with AI, including Nvidia (NVDA), AMD (AMD) and Alphabet (GOOGL). However, the biggest weight in the ETF is actually Meta Platforms (META) at 8.5%. Interesting, counter to the index construction rules listed above, META’s weight exceeds 8%, and the aggregate weight of positions with weights greater than 4.5% is actually 46.1%, so investors should expect some redistribution of the index weights in December.
Returns
Since the change to the PSJ ETF’s underlying index was only made as of August 25, 2023, the IGPT’s historical performance is not representative of the current index methodology (Figure 4).
Figure 4 – IGPT index change made in August (invesco.com)
Instead, readers may want to consider the performance of the Nexgen Software Index as a guide for potential performance of the IGPT ETF.
Figure 5 shows the historical performance of the Nexgen Software Index. It has impressively delivered 1-year returns of 46.1% to October 31, 2023. However, longer-term returns have been more modest, with 3 and 5-year average annual returns of 8.0% and 15.2% respectively.
Figure 5 – NexGen Software Index returns (stoxx.com)
Portfolio Construction Looks Fine, But Beware Of FOMO
Looking at IGPT’s portfolio, it looks appropriate for the AI theme as it contains several major AI-related companies such as Nvidia, Alphabet, and Meta. However, there are several concerns that I have with IGPT.
First, having firsthand go through designing ETFs, I can confidently say that many of these thematic ETF’s underlying indices are often ‘backed into’ by the index provider. In other words, the fund sponsor has a list of stocks it wants to include, and the index provider is tasked to come up with selection rules that will ‘produce’ an index holding these stocks. What these rules do not do is look into the future at potential disruptors and new entrants.
For example, OpenAI, one of the leading pioneers of AI, is currently a private company. For obvious reasons (i.e. private company does not have public shares), OpenAI is not included in IGPT. However, it is unclear if OpenAI were to seek an IPO and become public, would it qualify for the IGPT given the subsector and market cap criteria listed above.
Furthermore, there are many leading AI businesses/research centers within traditional companies that are not included in IGPT such as Tesla (TSLA) with its Tesla Bot and FSD driving software; GM (GM) which owns Cruise, one of the leading autonomous driving pioneers; and Hyundai (OTCPK:HYMTF) which owns Boston Dynamics, a leading robotics company.
In fact, AI and robotics is such a broad theme that it is hard to narrow it down to merely 100 stocks, since almost all companies are actively working on automation and software to some degree.
Finally, by jumping on these thematic ETFs, investors run the risk of chasing past performance. Just recently, I came across an article in the Financial Times discussing research from Morningstar that propose average investors in thematic funds “miss out on more than two-thirds of their potential total returns over the past five years due to buying high and selling low”.
On average, thematic funds returned 7.3% annualized to June 30, 2023, but the typical investor only received 2.4% per year, as many bought in after the bulk of the returns have been made (Figure 6).
Figure 6 – Investors tend to chase past performance in thematic funds (ft.com)
For example, while the NexGen Software Index has very impressive 1 year returns of 46.1%, the index returns have been much less impressive in recent months as the market digest the large rallies in the AI-related stocks from earlier in the year (Figure 7).
Figure 7 – NexGen Software Index has been muted in recent months (stoxx.com)
Conclusion
The Invesco AI and NexGen Software ETF was recently renamed and refocused on the ultra-hot AI theme. Looking at IGPT’s portfolio, it appears appropriately constructed to include many of the leading AI-rated stocks admire NVDA and GOOGL. Investors who want to track or gain exposure to the AI theme can consider the IGPT ETF.
However, investors are cautioned that IGPT’s index construction methodology may miss AI developments in many traditional companies admire GM and Tesla. Furthermore, by chasing thematic investments, investors may be at risk of buying high and selling low.
I rate the IGPT ETF a hold.
