About a third of the US convertible bond market is now linked to the AI trade, on Bank of America's own count, because that is where the hyperscalers, the neo-clouds and the data center companies have gone to fund their spending.
The usual way to own the AI build-out is the common stock, with the full drawdown that comes with it. Michael Youngworth's case is that the same theme can be held through the bonds financing it, and that the trade-off is a smaller share of the upside in return for a smaller share of the fall.
"You're not eliminating exposure to the AI theme. You are changing the shape of that exposure."
Youngworth runs global convertibles and preferred strategy at BofA Securities, and the drawdown, valuation and fund-flow figures in this episode are his own desk's.
The full episode is covered here so you can skip it.
Here are the 7 arguments that matter.
👤 Speaker: Michael Youngworth, Head of Global Convertibles and Preferred Strategy at BofA Securities
📰 Published: 14 September 2026 on YouTube (Bank of America) · recorded 8 September 2026
🔴 YouTube | 🔗 Episode page | ⏱️ 10 min
Key Takeaways
A convertible bond is a corporate bond and an equity option bought together
The bond pays income and cushions the fall, the option captures part of the stock's rise
About a third of the US convertible market is now connected to the AI trade
Alphabet, Oracle, CoreWeave, Nebius, Cipher and TeraWulf have all used convertible securities to fund AI spending
US convertibles fell 11% at their worst over two years while BofA's AI baskets fell 30–40%
The correlation to the same AI themes was about 80%
Convertible investors want volatility, because the option inside the bond is worth more when the stock moves
On two-thirds of US convertible issuers the stock has been moving more than convertible buyers paid for
Convertible issuance is on course for a record year, and the issuers are twice the size they were
The median US issuer's equity market cap is about double the level of three years earlier
July was the convertible market's worst month since 2022, and it is why the entry point is better now
1. What a Convertible Is
The episode opens with a definition, because Youngworth's argument depends on a structure most equity investors do not hold. A convertible bond starts as debt and carries the right to swap into the issuer's stock, which means the holder owns two things at once.
"The easiest way to think about it is two investments packaged together: a traditional corporate bond plus an equity option or a warrant." The bond provides income and some downside support; the option lets the holder participate if the stock rises
"That combination creates an asymmetric return profile. You can capture a meaningful portion of the stock's upside while typically experiencing less of the downside."
He is explicit that the cushion is not a floor: "Still, converts are a high-beta risk asset, and the floor is not a guarantee." If the issuing company runs into serious trouble, he said, both the stock and the convertible can fall sharply
Under more normal market conditions, the bond component can provide a cushion common stock does not have
On what the holder pays for that: "In practical terms, you're giving up some upside in exchange for a potentially smoother ride."
The structure has been rewarded this year: "Year-to-date, both US and global convertible bonds outperform broader equity and fixed income markets on both absolute and risk-adjusted bases, underscoring the benefit of the structure's asymmetric exposure to underlying common shares."
He does not leave the performance there — the structure alone, he said, does not explain this year's gains, and the rest of the answer is the asset class's exposure to AI
2. Converts Fund the Buildout
The second argument is about who is issuing. AI spending on data centers, compute, memory, power and the surrounding infrastructure has to be paid for, and Youngworth says the convertible market has become one of the places it is being paid for.
"That spending has to be financed. Increasingly, it is being financed in the convertible bond market."
The issuers he named run across the whole AI supply chain: hyperscalers Alphabet and Oracle, neo-clouds CoreWeave and Nebius, and data center companies Cipher and TeraWulf have all used convertible securities to help fund AI-related growth and capital spending
The reason an issuer chooses this over a straight bond is price: "Because the investor receives an equity option, the company can borrow at a lower coupon than it would pay on traditional debt."
That makes converts attractive capital for companies that need to spend heavily today in pursuit of growth tomorrow
The result, on his numbers, is that approximately a third of the US convertible bond market, or about 30% of the global market, can now be linked to the AI trade
"Converts are no longer merely adjacent to AI as they become one of the financing engines behind the AI buildout."
3. Timing Matters Less
The third argument is the one the episode is built around, and it is aimed at an investor who already believes the AI story. Youngworth frames the problem as ownership rather than conviction.
"While managers often recognize the transformational potential of AI, the harder question is how to own it." AI-related equities have delivered extraordinary returns, he said, but have also been prone to sharp drawdowns
Timing matters enormously when the underlying stock can move 20, 30% or even more in a relatively short period
The comparison he puts behind that is his own desk's: "Over the last 2 years, several of our AI-focused BofA custom baskets, including those tied to compute and AI infrastructure, saw a max drawdown of approaching 30% to 40%, while US convertibles experienced the max drawdown of just 11% despite maintaining meaningful correlation of about 80% to the same AI themes."
The broad equity indices did not do as well on the same test: he said the S&P 500 and the Nasdaq showed larger drawdowns of 19 and 20% over the same time frame, despite similar or weaker correlations to the AI thematic baskets
"This is the key point. You're not eliminating exposure to the AI theme. You are changing the shape of that exposure."
What the structure buys, in his framing, is meaningful participation with less of the path risk an investor has to tolerate along the way
4. Volatility Becomes an Asset
The fourth argument is the one he calls counterintuitive, and it is the part of the case that has nothing to do with the direction of the stock.
"Equity investors normally treat volatility as a problem. Convertible investors treat it as an asset." An option becomes more valuable when the underlying stock becomes more volatile, because the stock has a greater chance of moving far enough above the conversion price for the option to be worth something
He put a limit on that immediately: a market crash is not good for converts, and if stocks fall sharply enough convertibles will fall too
What helps is elevated single-stock volatility and wide dispersion between individual names, which is when the embedded option becomes more valuable
"That is particularly relevant in AI where index vol can look relatively calm even as the individual stocks move dramatically in opposite directions."
On where single-stock volatility sits now: "Average single stock vol among US convertible issuers now stands in the 97th percentile of its range in 2021."
The pricing conclusion is that buyers have been getting more movement than they paid for: "So in plain English, the market has been frequently delivering more stock movement than the convertible investors have been paying for." He said roughly two-thirds of US convertible issuers were in that position
The last piece is tenor. A typical convertible has a maturity of around five years, which he said can give investors long-duration optionality that is difficult or expensive to replicate in the conventional options market — particularly in smaller AI infrastructure, data center and alternative energy names where the listed stock options are not especially liquid
5. A Record Year of Supply
The fifth argument goes after a second assumption: that convertibles are financing of last resort for small speculative growth companies. He concedes there was historically some truth to it and says it is much less accurate today.
"Year-to-date, global convertible issuance has reached about 175 billion and is running at an annualized pace for over 250 billion, which would represent a record by a wide margin."
On the size of the market that supply is landing in: "The global convertible market as a whole has now grown to approximately 600 billion, including nearly 400 billion in the US."
The quality argument is in the median issuer: "The median US convertible issuer has an equity market cap of about 6 billion, roughly double the level 3 years earlier."
"Investment-grade issuers now represent approximately 20% of the global market, the highest share of the post-pandemic period."
The conclusion he draws is about who is in the market rather than how big it is: converts are no longer simply a niche financing market for companies that have exhausted every option, and now include larger, better-known names funding major strategic investments
6. The Entry Point Improved
The sixth argument is about when, and it rests on the summer's AI-led sell-off having hurt the asset class less than the stocks behind it.
"In July alone, US converts fell more than 5% during the month, their largest monthly decline since 2022. But the underlying stocks declined at least 7%, again illustrating the more defensive behavior of the convertible structure."
That sell-off is what made the valuation case: model-implied valuations cheapened to their most attractive levels of the year so far
He qualified it rather than selling it: "To be clear, converts are not particularly cheap versus history. Valuations are still around the 70th percentile over the previous 10 years."
Concentration in the largest issuers has also declined from its recent peak
On the demand side he wants improvement without crowding: "US convertible retail funds have attracted over 2 billion of net inflows during the last six weeks, equal to approximately 7% of assets." That marks the strongest six-month flow period since mid-2021 on his figures
Cumulative fund flows over a longer horizon, he said, suggest the asset class remains under-owned relative to its history: "Interest is returning, but investors do not appear to have fully embraced the product."
7. Three Ways to Own AI
The close sets the three options side by side rather than arguing that one of them wins outright.
Own the common stock and accept full upside and downside
Own traditional corporate debt, collect the income, and get limited participation if the equity story works
Look at the middle ground — convertible bonds, which he says offer equity participation, downside support from the bond component, and exposure to the volatility that becomes more valuable when individual stocks move sharply
He does not finish on a recommendation: "They're not risk-free. They are not simple substitutes for either stocks or bonds, and the structure and terms of each individual security matters."
The investor he is addressing is the one who believes in the long-term AI build-out and is less confident about timing every rally and every correction: "You can stay in the trade, you can participate in the upside, and you may not have to absorb every bump along the way."
Bonus Insights
The episode is a single unbroken monologue with no host and no questions. Signals & Noise is described on air as the place where strategists and economists from around the globe offer a shorter take on markets and economic matters as part of global research at BofA
Youngworth names his audience in the first minute rather than pitching the asset class in general: "If you believe in the long-term AI story, but you are increasingly uncomfortable with volatility, valuations, or simply the risk of getting the timing wrong, there may be a part of the market worth paying attention to: convertible bonds."
He is blunt about the asset class's reputation: "Converts are often dismissed as too complicated, too niche, or simply an obscure corner of the financial market. But that view is increasingly out of date."
He recorded the episode on 8 September, six days before it was published
Youngworth's bottom line is that an investor who believes in the AI build-out but not in their own timing can hold the theme through the convertible bonds financing it, keeping most of the correlation to AI while taking a fraction of the drawdown.
Products, Companies & Tools Mentioned
BofA Global Research (The source of the AI-focused custom baskets, the drawdown comparison, the model-implied valuations and the issuance and flow figures)
Alphabet and Oracle (The hyperscalers he names as having used convertible securities to help fund AI-related growth and capital spending)
CoreWeave and Nebius (The neo-clouds on the same list of convertible issuers financing the AI build-out)
Cipher and TeraWulf (The data center companies he names alongside them)
S&P 500 and Nasdaq (The broad equity indices in his two-year test, which he says showed larger drawdowns of 19 and 20% despite similar or weaker correlations to the AI baskets)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: