UBS On-Air · · 35 min

How should I be positioned? with Eric Van Nostrand (Lazard) and Jason Draho (UBS CIO)

With Eric Van Nostrand, Jason Draho

Summary

Eric Van Nostrand of Lazard and Jason Draho, UBS's CIO, debated how investors should position as rates, fiscal policy and AI-driven growth interact. Van Nostrand argues that the drivers of long-end yields (Fed policy, inflation expectations and term premium) are all pushing rates higher, that the US fiscal path leaves only a bond-market rejection or a crisis as the likely catalyst for policy change, and that near-term data center capex will slow amid returns scrutiny and anti-data-center politics in some states. He favors quality stocks with durable moats, including software names whose advantages lie in proprietary data and network effects, and emerging markets that have missed the AI enthusiasm. Draho sees the 10-year (about 4.88% at recording) near the low end of a persistent range, driven by nominal GDP of roughly 5% to 6%, and caps the 2-year at about 5.3% to 5.5%. He thinks financial repression is a live tail risk, which is why he prefers equities to fixed income. He views the data center policy reaction as overdone, since 2027 investment is already funded, and argues that capital, energy, chips and compute demand are bigger bottlenecks than regulation.

Key takeaways

  • Eric Van Nostrand on US Treasuries & yields: Rates have not peaked; inflation, Fed expectations and term premium all point to higher long-end yields.
  • Jason Draho on US Treasuries & yields: The 10-year yield should trade in a 4.9%–5.3% range, near the low end now, with 4.9% as year-end 2026 forecast.
  • Eric Van Nostrand on US deficit & government debt: US fiscal arithmetic leaves no politically viable fix, so a bond-market crisis may be needed before policymakers act.
  • Eric Van Nostrand on AI: overall outlook: AI's productivity shock, not near-term capex, will drive asset prices over the next decade.
  • Jason Draho on AI infrastructure & data centers: Market is too pessimistic on data center curbs; 2027 build-out is funded, and delays mainly hit 2028 and beyond.

What was said, by company and theme

13 views from 2 speakers. Every quote is checked against the episode's transcript (how).

US Treasuries & yields

Where everyone stands on US Treasuries & yields

  • Eric Van Nostrand · Lazard Asset Management

    Rates have not peaked; inflation, Fed expectations and term premium all point to higher long-end yields.

    He argues the three main drivers of interest rates (expected Fed policy, expected inflation and the term premium investors need for long-dated debt) are all pushing toward higher rates for the first time in most investors' careers. Central banks are being pushed toward tighter policy by rising inflation expectations, and supply constraints add a structural inflation impulse the Fed cannot control in the near term. Worries about fiscal sustainability are making long-horizon borrowing more expensive. He expects none of these drivers to stop, so long-end rates should stay higher for longer. He is more worried than Draho about the near-term rate path, because the pressure could compound and worsen before any policy response.

    “Short answer is no, we don't think they peaked. We think they have some room to run.”at 1:41
  • Jason Draho · UBS Global Wealth Management

    The 10-year yield should trade in a 4.9%–5.3% range, near the low end now, with 4.9% as year-end 2026 forecast.

    He says the 10-year, around 4.88% when recorded and down from about 5.3%, is now at the low end of a range that will likely persist. Rates have risen this year mainly because growth has been strong, with nominal GDP running at roughly 5% to 6%, which he expects to continue. He says the 2-year can't go much above 5.3% to 5.5% because that would require 3% growth indefinitely and a neutral fed funds rate near 4.5% to 5%, which few people accept. Going higher would mean the Fed needs a recession to bring inflation down. The 10-year could easily return to 5.2% to 5.3%, but he doesn't expect it to fall much lower.

    “using the ten year, from 4.9, that's our official forecast for the end of twenty twenty six and could easily go back to 5.2%, 5.3%.”at 5:00
  • Jason Draho · UBS Global Wealth Management

    Financial repression is on the table as a policy response to high long yields, though not highly probable.

    He says many investors conclude that the only way to solve unsustainable US fiscal arithmetic is financial repression, meaning policy to suppress rates. He points to the Fed not hiking in July after hawkish signals, and the Treasury's August announcement of small long-bond buybacks, which he reads as a baby step toward repression. He suggests regulation could push banks to hold more Treasuries. He notes the Bank of Japan's yield curve control as a precedent and says the US may follow that path later, but he doesn't see a high probability of it.

    “We may follow that path later on. ... Don't think it's a high probability. I think it's on the table of development possibility.”at 18:23

AI infrastructure & data centers

Where everyone stands on AI infrastructure & data centers

  • Eric Van Nostrand · Lazard Asset Management

    Data center build-out will slow near term, with forecasts disappointing and equities tied to it at risk.

    He expects data center investment to keep growing but to slow over the next couple of months, for two reasons. Businesses are becoming more risk-aware about the returns on incremental investment, and governors with explicitly anti-data-center positions gained power in the midterms. He expects this to disappoint current sky-high forecasts for new builds and to create risk for equities heavily tied to that spending. He also says AI investment is crowding out traditional capex and structures, which has been an important source of US growth. He thinks the market is overstating the political impact: governors will talk about moratoria but look for ways to let investment proceed, and local opposition will be smaller than feared. He still sees this as a near-term slowdown, not a long-run retreat from AI investment.

    “And as enthusiasm for building more and more data centers slows, it's not going to stop, but I do think it's going to slow both because businesses are starting to get a little more risk aware of the returns to incremental investment from here.”at 22:46
  • Jason Draho · UBS Global Wealth Management

    Capital access, energy, chips or weak compute demand are likelier than data center rules to slow AI investment.

    He argues that other bottlenecks are more likely than data center regulation to slow AI investment. These include constrained access to capital, energy supply, semiconductors, and compute demand that falls short of what is being built. He says data center regulation would have at most a marginal impact, so it is not a game changer, while these other constraints could matter more.

    “there are other potential bottlenecks that are maybe more likely to cause a slowdown in AI investment, whether it's access to capital gets constrained and access to energy supplies, semiconductors, or just the demand for compute is not nearly as significant as what is being developed.”at 27:59
  • Jason Draho · UBS Global Wealth Management

    Market is too pessimistic on data center curbs; 2027 build-out is funded, and delays mainly hit 2028 and beyond.

    He sees the market reaction to tighter data center policy as overdone. The Texas approach he expects supports AI data centers if they bring community benefits and manage electricity demand, water use and grid constraints. Projects postponed or put under moratoria would hit investment in 2028 or later, while 2027 investment is already permitted and funded. As a result, he thinks the investment-driven growth and earnings story is largely done, and the market now cares about 2028 and beyond.

    “anything that sort of is postponed, delayed, or moratorium probably impacts investment in 2028 or beyond. Everything for 2027 is almost like that's permitted, there's capital.”at 29:01

AI: overall outlook

Where everyone stands on AI: overall outlook

  • Jason Draho · UBS Global Wealth Management

    AI investment lifts rates near term, but the aim is a productivity boom like the 1990s internet era.

    He describes a feedback loop. Strong AI investment over the next one to two years has driven growth and higher rates, so rates are partly endogenous to the AI story. The longer-term goal is a shift to higher productivity growth, which would let the economy grow faster without generating inflation. That would make higher rates a good reason rather than a bad one. He notes that the Trump administration and Kevin Warsh have framed this as a 1990s-style productivity boom, which he hopes will repeat with AI as the internet did. The argument rests on hope and the historical analogy, not hard data.

    “But ultimately, the aim would be this transition to higher productivity growth. And if you have higher productivity growth, well, then you can also grow faster without generating inflation. It may mean rates are higher, but it's kind of the good reason why rates are higher.”at 19:08
  • Eric Van Nostrand · Lazard Asset Management

    AI's productivity shock, not near-term capex, will drive asset prices over the next decade.

    He separates AI's demand effects (data center construction, frontier lab profitability), which matter over the next couple of months, from its supply-side productivity effect, which he expects to be the bigger source of asset-price variance over the next decade. Model weights are easily transferred, and open-weight Chinese models show they can be tuned quickly, so the benefit should spread. Even in pessimistic scenarios for data center investment over the next year or two, he thinks the economy is at the inflection point before a historic upside productivity shock. He is more pessimistic about near-term GDP impulse than about long-run supply-side growth, and points to how quickly model capabilities have improved as evidence.

    “But to my mind, the bigger source of variance in asset prices over the next decade as opposed to the next couple months is gonna come from the supply side components of AI, which is the productivity shock.”at 20:37

US stock market outlook

Where everyone stands on US stock market outlook

  • Jason Draho · UBS Global Wealth Management

    Equities are preferred over fixed income because they benefit from higher nominal GDP, especially if repression risk grows.

    If there is even some chance of financial repression that keeps rates lower, he doesn't want to own fixed income. He wants assets that benefit from higher nominal GDP, such as equities. With nominal GDP running around 5% to 6%, equities are better placed to benefit from the growth backdrop that is driving rates higher.

    “I don't necessarily want to own fixed income, I want to own assets that would benefit from higher nominal GDP like equities.”at 29:50
  • Eric Van Nostrand · Lazard Asset Management

    Quality companies with moats and balance sheets should outperform as AI-driven exuberance fades.

    Lazard's equity approach has long focused on quality companies with strong balance sheets, which earn their returns through persistent earnings-per-share growth from competitive moats rather than from timing exuberance around a technology or sector. Many companies with those traits have been beaten down this year as AI enthusiasm rose. He thinks the quality factor is one area where meaningful outperformance could show up as the near-term AI demand story cools.

    “We identify companies with the kind of strong balance sheets that aren't getting their equity market returns from trying to time exuberant sentiment on any particular technology or sector, but that have strong competitive moats to deliver persistent earnings per share growth over time.”at 30:47

US deficit & government debt

Where everyone stands on US deficit & government debt

  • Eric Van Nostrand · Lazard Asset Management

    US fiscal arithmetic leaves no politically viable fix, so a bond-market crisis may be needed before policymakers act.

    Drawing on his time at Treasury, he argues the rise in interest expense over the next decade leaves only three ways to get to a sustainable fiscal path: raise taxes on the middle class, cut Social Security and Medicare, or get a bailout from long-term rates. He says the first two are politically impossible. The third fails because the Trump Treasury's attempts to treat the 10-year yield as a policy instrument, including small buybacks of long-dated bonds, have not moved markets, and he doubts the Fed can control the 10-year enough to change the fiscal picture. That leaves a fourth option, a clear bond-market rejection of the status quo. He is pessimistic about global fixed income for this reason and expects policymakers to act only after a crisis-level spike in yields.

    “I'm worried that it's going to take a crisis level spike in bond yields before policymakers do anything about the issue. And that's what's got me nervous about fixed income right now.”at 17:42

Software & SaaS

Where everyone stands on Software & SaaS

  • Eric Van Nostrand · Lazard Asset Management

    Beaten-down software names with data and network-effect moats are poised for a rebound.

    Software names have been hit hard this year as AI enthusiasm rose. He argues that the old software moat, having the best coders, is weak now that AI models have matched that scale. Companies whose moats come from proprietary data, network effects and being the industry-standard provider are different. Those moats were never in the code, so he thinks many of these names are poised for a rebound. He links this to Lazard's quality approach, which favours companies with real barriers to competition, and says the pullback in near-term AI investment is well placed to benefit them.

    “Those moats were never in the code. Those moats were in the data and in the network effects, and the barriers to competition. I think those, a lot of those names are poised to see an important rebound.”at 32:21

International equities

Where everyone stands on International equities

  • Eric Van Nostrand · Lazard Asset Management

    Emerging markets offer a near-term way to benefit from the supply-side story, with little AI enthusiasm priced in.

    He argues that investors can benefit from the long-term supply-side AI story in the shorter term by looking beyond the large rich economies to emerging markets. With some exceptions, emerging markets have not received the AI enthusiasm that companies tied to US frontier labs have. Because that enthusiasm is largely absent, emerging markets are a place to find value and exposure to the productivity story, and he calls this an important area to push into.

    “The emerging world... is a place where, with some obvious exceptions, but in general hasn't gotten the gusto of the AI enthusiasm in the way that companies adjacent to American frontier labs have.”at 32:55

Topics

  • 10-year yield outlook
  • AI investment bottlenecks
  • AI productivity boom
  • AI productivity shock
  • data center buildout policy risk
  • data center capex
  • emerging market equities
  • equities vs bonds allocation
  • financial repression
  • fiscal sustainability and bond market
  • long-end yields path
  • quality factor
  • software valuations under AI

More from UBS On-Air

All UBS On-Air episodes

Ask your AI about this episode

Chorus gives Claude, ChatGPT and other assistants every view from this episode and every other show Chorus follows. Ask who agrees, who changed their mind, and what else was said about the same companies.

Setup guide

A weekly email with the calls that mattered from the most important voices in investing. Unsubscribe any time. Privacy Policy

Chorus reports what speakers said. It isn't investment advice. How Chorus works · About Chorus