1. Markets have spent the last two years debating whether AI changes productivity. I think the next two years will be spent discovering whether it changes the neutral rate (r-star). That is ultimately the more important question because it sits behind almost every macro debate that matters today: inflation, the cost of capital, valuations, the dollar and, ultimately, who captures the economics of AI. The events of the past fortnight have reinforced that view rather than challenged it. Markets have understandably focused on reports of an Iran-U.S. understanding around Hormuz and the subsequent collapse in oil, with Brent now trading back below pre-conflict levels. The instinct has been to conclude that we are moving back towards the benign disinflationary regime investors became accustomed to over the last decade. I think that is the wrong takeaway. Lower oil changes the timing of inflation. It does not change the regime. I think markets are asking the wrong question. AI is no longer simply changing productivity. It may be changing r-star itself.

2. Markets still think about AI predominantly through an equity lens. I think it should increasingly be viewed through a rates lens. AI capex expectations have moved from roughly $400-500bn a year ago to around $750-800bn today, with consensus approaching $1tn and some estimates already above that for 2027. From a macro perspective, those numbers tell a very different story. They suggest demand for capital is rising faster than the economy can supply it. Before AI delivers meaningful productivity gains, it requires enormous investment in power generation, transmission networks, semiconductor fabs, cooling systems and data centers. Those bottlenecks support investment, wages and pricing power long before they improve productivity. AI is inflationary first and disinflationary later. Markets continue to price the second half of that story far more aggressively than the first.

3. Which brings me to what I now think is the more important debate. Markets remain focused on whether inflation returns to target. I think the bigger question is whether r-star itself has moved permanently higher. Inflation can moderate while equilibrium real rates remain structurally higher. If AI creates a sustained increase in investment demand while competing for scarce physical resources over many years, the economy may simply require a higher real rate than markets have become accustomed to since the GFC. Markets continue to assume that once inflation falls, rates naturally follow. That only holds if r-star itself is unchanged. I am no longer convinced that assumption is right.

4. This has probably been the biggest evolution in my own thinking over the past six months. Markets continue to analyze AI as another software cycle. I think that analogy is fundamentally wrong. Software lowered capital intensity. AI is doing almost the opposite. It requires power stations, transmission networks, semiconductor fabs, cooling systems, data centers and hundreds of billions of dollars of physical investment before the productivity benefits ever fully materialize. Software cycles tend to lower equilibrium rates because they require relatively little capital. Infrastructure booms tend to raise them because they compete aggressively for scarce capital. We keep describing AI as a software cycle while building it like a railway. That distinction may prove to be one of the defining macro shifts of this cycle.

5. The biggest risk to this framework is also the one I spend the most time thinking about. If AI ultimately delivers the productivity gains its champions expect, today’s bottlenecks may prove to be transition costs rather than a permanent claim on capital. Falling inference costs, open-source models and rapidly improving model efficiency all point in that direction. I still favor the higher-for-longer interpretation because the investment is happening today while the productivity remains largely a promise. That does not mean the productivity never arrives. It simply means markets may be bringing forward the benefits while underestimating the cost of getting there. Timing matters in macro, and for now the investment cycle is arriving well ahead of the productivity cycle.

6. None of that materially changes my policy view. I still think the next move is more likely to be a hike than a cut and that this cycle ultimately ends through the cost of capital rather than through a collapse in demand. Where I have become more nuanced is around the path. Lower oil, falling gasoline prices and favorable base effects should produce softer headline inflation over the summer, and a negative monthly CPI print would not surprise me. I suspect markets will interpret that as evidence the inflation problem has largely been solved. I think that conclusion is premature. Headline inflation can soften materially while the underlying drivers of nominal growth remain intact. That is why I think July is less interesting than markets currently believe. OIS prices around 7-8bp into the meeting, implying roughly a one-in-three probability of a hike. My own estimate remains closer to 15-20%. Kevin Warsh, the new Fed Chair, did exactly what he needed to at his first meeting. He established credibility around price stability while revealing remarkably little about his reaction function. Waiting preserves optionality, keeps every subsequent meeting genuinely live and, in my view, makes considerably more sense than rushing into an early hike simply because markets briefly give him permission. If July passes without action, which remains my base case, relatively little changes beyond some uncertainty premium coming out of the front end. My disagreement with the market is therefore more about distribution than destination. I continue to think policy ultimately settles tighter than consensus expects. I simply think the journey there is likely to be slower and more patient than current pricing occasionally implies.

7. The dollar has quietly become one of the clearest confirmations that the regime is changing. Historically, a 20% decline in oil would have supported Asian FX and commodity currencies through the terms-of-trade channel. Instead, that relationship has steadily weakened as relative policy expectations have become the dominant driver. FX is once again trading rates rather than commodities. A more credible Fed, less forward guidance and an AI-led investment cycle consistent with a higher equilibrium rate all point in the same direction. DXY strengthening despite materially weaker oil is therefore less a contradiction than confirmation that markets are repricing the cost of capital rather than the price of energy. It also helps explain why Asian FX has struggled to rally despite a meaningful improvement in its energy terms of trade.

8. The same framework also helps explain what has happened beneath the surface of equity markets. The AI trade has fractured rather than broken. Memory manufacturers, networking companies and other infrastructure beneficiaries have continued to outperform, while several hyperscalers have lagged, rewarding those supplying the build while becoming more selective towards those funding it. For more than a decade, the largest technology companies provided a structural source of equity demand through aggressive buybacks. Today, a growing share of that capital is being redirected towards AI infrastructure. Friday’s reaction probably told us more than Micron’s results themselves. Blowout numbers were followed by a sharp sell-off across semiconductors, while hedge funds sold U.S. equities at the fastest pace since Liberation Day despite already elevated net exposure. That looked far more like a funding and positioning adjustment than a reassessment of AI itself. As investment requirements continue to rise, I suspect markets will become progressively more focused on who can finance the build than simply who builds the best models.

9. The funding story remains, in my view, one of the most underappreciated parts of the current setup. Markets continue to focus almost exclusively on central banks while overlooking the mechanism through which financial conditions are already tightening. Prime broker balance sheets have quietly become one of the most important transmission channels in the system. Financing costs, balance sheet allocation and collateral constraints are delivering part of the tightening that monetary policy has yet to produce. Ownership remains heavily concentrated, while many of the same crowded positions continue to rely on the same balance sheets for financing. Asian assets continue to consume scarce unsecured balance sheet, allowing local funding pressures to spill into US and European markets more quickly than fundamentals alone would suggest. Over shorter horizons, funding conditions matter as much as valuations. If tighter monetary policy ultimately collides with tighter funding, repo becomes the policy rate that matters at the margin. The lesson from 2019 was not simply that reserves matter. It was that funding markets can tighten financial conditions much faster than traditional macro models assume.

10. Pulling it all together, I do not think the medium-term regime has changed very much. Lower oil gives central banks greater tactical flexibility over the coming months, but it does very little to alter the strategic backdrop. AI is no longer simply a technology story. It is a capital allocation story, an inflation story and, ultimately, a rates story. The same bottlenecks supporting earnings today are also supporting nominal growth and lifting the equilibrium cost of capital. Markets have spent the last two years debating whether AI changes productivity. I think that was the easier question. The harder question is whether AI changes r-star itself. If that framework is broadly right, inflation, the dollar, funding conditions, nominal growth and valuations cease to be separate debates. They become different expressions of the same underlying variable: the price of capital. Identifying AI as the investment theme was the easier trade. Understanding who ultimately captures the economics, and how markets finance the build, is likely to be the harder one. If AI is changing the price of capital rather than simply productivity, investors may be materially underestimating where equilibrium rates ultimately settle. I think that will prove to be the defining macro question of the next two years.

Macro Commentary Disclosure

The views expressed herein are those of Schonfeld Strategic Advisors LLC (“Schonfeld”) as of June 29, 2026, and are provided solely for informational and discussion purposes. These comments represent our current judgment based on information available at the time of writing and are considered “forward-looking statements” subject to change at any time without notice. Certain information contained herein has been obtained from publicly available sources believed to be reliable; however, no representation or warranty, express or implied, is made as to its accuracy, completeness, or timeliness. This commentary should not be construed as investment advice nor a recommendation to buy or sell any security, strategy or fund, and any references to specific instruments, trade structures, or market expressions are for illustrative purposes only and do not reflect actual or intended positioning by Schonfeld or any of its funds. This commentary is directed to sophisticated market participants and does not purport to define or explain all industry-specific terminology contained herein. Future results may differ materially from expectations due to evolving market conditions, new information or subsequent events. This commentary is intended as a general macro or strategy-level discussion and does not replace fund-specific disclosure materials. The views expressed do not guarantee the future performance of any security, asset class or market and should not be relied upon as predictive of any Schonfeld account, portfolio or fund performance. The content included herein is confidential and may not be reproduced or distributed, in whole or in part, without the prior consent of Schonfeld.