Great Rotation Happening Now: This Sector ‘Is Over’ | Thomas Hayes

Comstock Resources

• Thomas J. Hayes · Great Hill CapitalBullish

  Comstock is the lowest-cost gas producer near Gulf export terminals and AI data centers, worth $40-60.

  Hayes presents Comstock as an AI picks-and-shovels play. He says it is the lowest-cost natural gas producer in the country by 50%. It trades at about $3.7-3.8 billion market value at $13-14 a share. Its proved and unproved reserves are estimated at about $20 billion. Its Haynesville Shale acreage in Western Louisiana sits close to Gulf export terminals and to the East Texas data centers, and it is part of a project tied to Korean investment. He says the stock can be a $40-60 stock over three to five years, though he notes it is an opinion, not advice, and that is how his firm underwrites it.

  “Number one, we own a company called Comstock Resources. They are the lowest cost natural gas producer in the country by 50%.”at 9:08

Alibaba Group Holding

Across the shows: 1 speaker in other episodes, 1 bullish and 0 bearish calls, past 30 days

• Thomas J. Hayes · Great Hill CapitalBullish

  Alibaba is the cheapest AI player; its unpriced stakes in AI startups like Moonshot are undervalued.

  Hayes says Alibaba owns equity in almost every AI startup in Asia. Its roughly one-third stake in Moonshot, which is reportedly going public at a $50 billion valuation, implies about $15 billion of equity value that is not priced into Alibaba stock, and that is one of dozens of such holdings that could list. He also cites Alibaba as the number one compute provider in Asia, with the lowest relative multiple, the healthiest balance sheet, and AI revenue growing triple digits for over ten quarters. He calls it a free call option on the Chinese middle-class recovery.

  “we have a company called Alibaba, which is the cheapest AI player in the world. They own equity in pretty much every AI startup in Asia. One of them is called Moonshot, which they own about a third”at 10:25

Crown Castle

• Thomas J. Hayes · Great Hill CapitalBullish

  Crown Castle wins because Starlink Mobile cannot work inside buildings without cell towers.

  Hayes says his firm brought Crown Castle to listeners before anyone wanted it. The market had feared that Starlink would disintermediate its towers and that its big customers, Verizon, AT&T and Sprint, would stop needing them. He argues that Starlink Mobile works in rural areas, outdoors and on boats or planes, but not for indoor coverage in buildings, where people need towers. When Musk bought about $8 billion of spectrum, the market recognized this and the stock rose 20% in a day. It also pays a 5.3% dividend yield, and he expects further upside.

  “the market figured out what we've been saying for months, which is that Starlink can't work. Starlink Mobile can't work inside buildings without self towers.”at 15:10

GXO Logistics

• Thomas J. Hayes · Great Hill CapitalBullish

  GXO is an AI beneficiary, using robots and humanoids to lift warehousing margins in the middle innings of AI.

  Hayes describes GXO as the largest pure-play warehousing and logistics company. He says it is already using robots, AI and humanoids in its warehouses to increase margins significantly. He sees companies that adopt AI, rather than build it, as the winners. He places GXO in the middle innings of the AI trade and says it will continue to do well.

  “AI beneficiaries that are going to utilize it, GXO Logistics, which is the largest pure play warehousing and logistics company. They're already using robots and AI and humanoids in their warehouses to dramatically increase margins.”at 17:27

QXO

• Thomas J. Hayes · Great Hill CapitalBullish

  QXO's Brad Jacobs roll-up will win as cheap housing-linked deals compound over three to five years.

  Hayes says his firm is betting on Brad Jacobs, whose roll-ups have a long record: United Rentals, XPO and Waste Connections. Jacobs is now building a roll-up of building supplies (roofing and installation), has done three large deals, and is heading toward $7.5 billion of EBITDA. The stock trades down because of about $4 billion of debt. Hayes says when housing stays down Jacobs does cheap deals because he raises money at high multiples and buys businesses at 5-10 times EBITDA, which are accretive. When housing recovers he sells products he already has. His firm is buying every share it can get over a three- to five-year view.

  “The number one way that we're playing housing is we're betting on a jockey called Brad Jacobs. Brad Jacobs has created more wealth for shareholders than any other operator in the history of mankind in roll ups.”at 22:28

AI infrastructure & data centers

• Thomas J. Hayes · Great Hill Capital

  AI debt financing rests on frontier labs' cash flows, which customers cannot yet produce at the needed scale.

  Hayes says about $500 billion of bonds are predicated on two frontier large language model companies meeting their income projections. Working backwards from those projections, OpenAI and Anthropic would need free cash flow about twice that of the other nine sectors of the economy combined. He says their customers currently generate only about half the free cash flow needed to make the model work. He concedes there is some froth but says AI itself is not over, so the risk is in which players end up losing.

  “if you estimate the amount of cash that OpenAI and Anthropic have to generate to make the math foot, the free cash flow is two times the free cash flow of the rest of the economy, the other nine sectors.”at 5:49

• Thomas J. Hayes · Great Hill Capital

  Companies investing heavily in AI capex historically end up as the biggest losers, not the winners.

  Hayes says history shows that companies investing in capital expenditure during technology booms are not the biggest winners, and more often they were the biggest losers. He argues the winners are the firms that implement the technology to raise productivity and margins. As evidence he notes Jeff Bezos is raising money to buy industrial manufacturing companies that will use AI to double margins rather than funding AI startups. He says everyone is chasing the commodity layer of semiconductors instead.

  “History would suggest, for anyone who's read a history book, is that those who are investing in the CapEx are not going to be the biggest winners. ... they were the biggest losers.”at 6:27

• Thomas J. Hayes · Great Hill Capital

  Hyperscaler CapEx spending will persist another five to seven years.

  Hayes argues the AI infrastructure cycle has years left. He says the CapEx spenders, the hyperscalers and the chip companies, will keep spending for another five to seven years. He asks how an investor can miss that and still outperform, which implies he sees the spending as durable and rewards staying exposed to the build-out.

  “The LLMs and the semiconductor trade and all this stuff and the hyperscalers, the CapEx spenders persist for another five to seven years.”at 41:13

• Thomas J. Hayes · Great Hill Capital

  Bullish on AI long term but not on big players debt-financing the buildout; benefits go to adopters.

  Hayes is bullish on AI's long-run effects on productivity, growth and prosperity. He is not bullish on the companies financing the buildout with heavy borrowing. In his view, the economic gains go to businesses that apply AI to their existing operations to lift margins and productivity, not to the firms taking on the debt to build capacity.

  “Bullish on AI, increasing productivity, increasing standard of living, increasing GDP growth, increasing prosperity in the long term, not bullish on ones financing all the build out, taking on all the debt, benefit on all the ones, applying it in their normalized businesses to increase margins and productivity.”at 42:52

US stock market outlook

• Thomas J. Hayes · Great Hill Capital

  Indices will stay subdued while a strong bull market builds under the surface in non-AI stocks.

  Hayes says that since July 31, 89% of S&P 500 stocks had fallen more than 10% and 54% had fallen more than 20% until about four days before the interview, yet the index was near all-time highs, held up by five or ten AI stocks. He reads this as a bear market under the surface with the indices holding up. He expects the reverse over the next six to twelve months: a large rally in the other 493 stocks, which carry lower index weights, while the index gains only modestly as the crowded AI leaders consolidate. He says the catalyst is yields stopping their rise and possibly falling.

  “I think over the next six months, you're going to have a monstrous bull market under the surface. And it's not going to be reflected in the indices.”at 13:17

• Thomas J. Hayes · Great Hill Capital

  Defensive sectors like utilities, staples and health care are being bid as money rotates from tech.

  Hayes argues that utilities, staples and health care, the predominantly defensive and rate-sensitive groups, have started to get bid in a material way over the last five sessions, and bonds are being bid too. He says investors can lock in a 5.3% yield on ten-year Treasuries versus a roughly 5% earnings yield on equities, and that pensions with about 6% hurdle rates are moving out of equities into bonds. He expects defensives such as Diageo, Baxter, health care and staples to work over the next six to twelve months, while the AI-linked leaders consolidate.

  “Utilities, staples, health care, which are the three predominantly defensive groups, and interest rate sensitive groups are starting to get bid in the last five sessions and bid in a material way.”at 18:14

• Thomas J. Hayes · Great Hill Capital

  Small caps will explode higher over nine to sixteen months as yields stop rising.

  Hayes says small caps just had their second-worst quarter relative to the S&P this century, while ISM manufacturing is in expansion for nine straight months and $8 trillion sits in money market funds, which is fuel for a broader rally. He says small caps are out of favor because they are seen as the most leveraged and borrow the most. He argues they are more nimble than large companies and can adopt AI faster for productivity and margins. As yields slow or fall, he expects small caps to rebound sharply over the next nine to sixteen months, with large operating leverage.

  “small caps are like a beach ball held underwater or a coiled spring, whatever analogy, ... that are going to absolutely explode higher, we believe, over the next nine to sixteen months.”at 32:55

AI chips & semiconductors

• Thomas J. Hayes · Great Hill Capital

  Commodity memory and chip trade is over short term as rivals add capacity and prices cure high prices.

  Hayes argues memory and commodity semiconductors are a classic boom-bust cycle. Chip margins were so high that China, Korea and Japan were bound to add capacity and flood the market, which is the 'high prices cure high prices' dynamic. He compares it to the earlier CPU-versus-GPU inference shift, when Intel was written off and later rose about five times. He says the story of semis and memory repeats every four years: a shortage, new capacity, competition arrives as demand falls, and stocks collapse. He had exited Intel and says semis fell 35-60% in July, with only a few names in a basket of 30 making marginal new highs before rolling over again. He concludes the commodity part of the AI trade is over in the short term.

  “why the hell is anyone surprised? I mean, it's a commodity. High prices cure high prices. Did you think the Chinese, the Koreans, and the Japanese were gonna sit back and say, oh, Micron. Look at those margins.”at 3:22

• Thomas J. Hayes · Great Hill Capital

  High memory prices will cure themselves as Asian producers flood the market with cheap memory and chips.

  Hayes says high prices cure high prices. The AI infrastructure providers' high prices will bring a flood of capacity from Chinese, Japanese and Korean producers. He argues the US cannot block these imports without losing the AI race, so the market will be swamped with cheap memory and semiconductors for a long time. He also points to DeepSeek's shift from GPUs to CPUs as a sign that cost-cutting will keep pressuring the chip trade.

  “High prices cure high prices. You're gonna see the market flooded, flooded with cheap memory and semiconductors till the cows come home.”at 41:52

US Treasuries & yields

• Thomas J. Hayes · Great Hill Capital

  Ten-year yield would fall to about 4.75% immediately if the Iran war ended.

  Hayes says yields are rising globally but in the US they are rising alongside real growth and nominal growth of about 8%. Part of the move reflects inflation expectations tied to elevated oil, with diesel a key watch point, and part reflects demand for capital. He says if the war ended tomorrow the ten-year would immediately fall to about 4.75%. After that, the level would depend on sustainable growth. If 8% nominal growth is not sustainable, he expects 4.25-4.50%. If it is sustainable, about 4.75-5%, but not the 5.5-5.6% people talk about.

  “if the war ended tomorrow, I think the ten year yield would be immediately down to four seventy five, and then we would start to figure out what level of growth is sustainable”at 26:50

• Thomas J. Hayes · Great Hill Capital

  A guaranteed 5.2% ten-year Treasury yield now competes well with the roughly 5% earnings yield on stocks.

  Hayes says the S&P trades at about 20 times earnings, which gives a 5% earnings yield. That yield is not guaranteed, though he says the earnings growth expectations are the highest since Q4 2021 and would have to be met. A 5.2% ten-year Treasury yield is guaranteed if held to maturity, with temporary mark-to-market risk if yields rise to 6%. He adds that the trade-off was not on the table six months ago but is now reasonable, especially for pension managers with about 6% hurdle rates, who will shift more of their portfolios from equities into bonds.

  “Alternatively, you can take a guaranteed 5.2% yield locked in for ten years with zero permanent downside.”at 29:56

AI bubble & tech valuations

• Paul Tudor Jones · Tudor Investment Corporation

  Market is bifurcated between AI adopters and non-adopters, moving in opposite directions, as in 1999.

  Jones argues the market is split into companies adopting AI and those that are not, with the two groups moving in opposite directions. He draws a direct parallel to 1999, when the same split occurred. He also cites a recent conference where attendees were far less experienced than they guessed, which he treats as a sign of the period's froth. The argument is largely qualitative, with the 1999 comparison as its main support.

  “You've got it completely perfectly bifurcated in the sense that you have obviously everything that's adopting AI and everything that's not, and they're going in opposite directions. And that was pretty much the case in 1999.”at 37:52

• Thomas J. Hayes · Great Hill Capital

  Semiconductors and memory are frothy and could be hit very hard, as the sector fell 50%+ three times in 1995-2000.

  Hayes says the 1995-2000 tech rally was parabolic, and during it the semiconductor basket corrected by 50% or more on three separate occasions. He uses that history to argue there is froth in certain pockets, especially semiconductors and memory, which could be hit very hard even while the broader AI theme keeps advancing. He presents this as a risk to crowded positions rather than a call that the AI trend is ending.

  “From 1995 to 2000, you had this parabolic move in tech. During that period, semiconductors as a basket corrected 50% or more, three separate discrete times. So I think there are froths. There is froth in certain pockets like semiconductors and memory that could be hit very, very hard while the theme continues forward.”at 38:58

AI: overall outlook

• Thomas J. Hayes · Great Hill Capital

  AI will still raise productivity and the market can keep rising; only the commodity chip trade is over.

  Hayes says that the commodity part of the AI trade is done, but that does not mean AI will stop raising productivity or that the market will stop going up because of what AI does for the economy. He argues AI is just getting started, and the key question is who wins and who loses. Winners will be those who use AI to raise productivity and margins, not those who fund its capex.

  “It doesn't mean that AI will not increase productivity. It doesn't mean the market won't keep going up because of what AI does for the economy and productivity.”at 8:52

China's economy

• Thomas J. Hayes · Great Hill Capital

  China's property market is bottoming and stimulus is returning, supporting a Chinese consumer recovery.

  Hayes argues that China's middle-class recovery is happening. Property prices, which have fallen for years, appear to have bottomed and started turning up for the first time in years. The government is back to stimulating, and he cites a stimulus package of about $80 billion announced last week. He uses this as a reason Alibaba, a 'toll taker' on the consumer, could benefit.

  “If you see property prices for the first time in years have bottomed and are starting to turn up in China, the government is back at stimulating. I think they did about $80,000,000,000 last week”at 10:52

International equities

• Thomas J. Hayes · Great Hill Capital

  International equities should outperform the US, as chip-led US leadership fades like in 2000-2006.

  Hayes says a broader non-consensus theme is emerging: international equities outperform the US, small caps beat large caps, value beats growth, and bonds get a bid. He compares it to 2000, when chips rolled over and the rest of the market, the 493 stocks outside the leaders, outperformed for years. His example is Alibaba, which he calls the cheapest AI player with the lowest relative multiple, healthy balance sheet and triple-digit AI revenue growth for over ten quarters, yet 'you can't give the stock away'. He expects that to change.

  “where international equities start to outperform The US, where small start to outperform large, where value starts to outperform growth”at 11:44

US housing

• Thomas J. Hayes · Great Hill Capital

  Housing is set to outperform over the long term given a 4 million home deficit and millennial demand.

  Hayes argues that the housing shortage fundamentals have not changed. The US still has a deficit of about 4 million homes. Millennials, now around age 35, are the largest population cohort and are forming families and households, so they need houses. He says that even if the near term is weak, housing will outperform over the long term. Any relief in rates would let homebuilders build smaller starter homes, which he compares to the 1950s boom.

  “Maybe it doesn't work in the next one to three quarters, but it's certainly going to outperform over the long term because nothing has changed. You still have a 4,000,000 house deficit.”at 21:08

Europe's economy

• Thomas J. Hayes · Great Hill Capital

  European bond yields are blowing out from fiscal irresponsibility and energy import dependence, as in France.

  Hayes distinguishes yield moves by region. Europe's yields are rising with no economic growth. He says Europe is picking up inflation relative to oil because it has no natural resources, so it is the most affected by energy prices, and it is fiscally irresponsible. He points to French bond yields blowing out as the consequence of fiscal problems. Japanese yields are rising for different reasons, with inflation at multi-decade highs.

  “In Europe, it's picking up relative to oil because they don't have any natural resources, so they're the most impacted and fiscally irresponsibility. France's bond yields now blowing out and the chicken's coming home to roost.”at 26:08

US consumer

• Thomas J. Hayes · Great Hill Capital

  Consumer discretionary is overshot to the downside; US spending will recover over twelve months.

  Hayes says consumer sentiment is at its worst level while retail spending is booming. He argues people 'hate spending' but still spend, citing restaurant reservations booked one or two weeks ahead on Friday and Saturday nights. He says restaurants, clothing and sneakers are in the gutter and overshot to the downside, and that the US consumer still has jobs with 4.2% unemployment and nominal growth near 8%. He expects a big recovery in those discretionary categories over the next twelve months.

  “Consumer discretionary is in the gutter, and I think that's overshot to the downside ... You can never bet against The US consumer when they have a job.”at 34:00

US growth & recession risk

• Thomas J. Hayes · Great Hill Capital

  The US is not heading into recession; jobs are still abundant and the economy keeps growing.

  Hayes says wages have risen somewhat and jobs are still abundant with unemployment at 4.2%. The economy is growing at about 8% nominal. He acknowledges the risk: the yield curve came within 28 basis points of inverting last month when the Fed hiked, and that could change if the Fed keeps hiking. For now, though, conditions are not recessionary, which supports a positive view of the real economy.

  “What they don't realize for the most part is wages have gone up a bit. Jobs are abundant still. That could change if the Fed keeps hiking, but it hasn't changed yet.”at 35:40

Enterprise AI adoption & AI software

• Thomas J. Hayes · Great Hill Capital

  AI beneficiaries deploying the tech for margins and productivity are early and offer asymmetric upside with protected downside.

  Hayes argues the best way to protect yourself if the AI trade unwinds is to buy AI beneficiaries rather than the capex spenders. These are companies that use AI to raise margins and productivity, not the companies building infrastructure. Because the technology is only in the first or second inning, he sees their downside as protected and their upside as asymmetric. Crowded capex trades, by contrast, carry a large downside if the cycle turns.

  “Buying the AI beneficiaries where your downside is protected and your upside is asymmetric, irrespective of whether the AI trade unwinds... The beneficiaries that are deploying this technology to increase margins and productivity, I think it's the first inning or second inning.”at 39:44

Open-source AI models

• Thomas J. Hayes · Great Hill Capital

  Businesses will route most tasks to open-weight models at half the token cost with 99% efficacy.

  Hayes expects spending to shift toward open-weight models. He describes orchestration tools that send each task to the cheapest capable model: for about 99% of tasks, a distilled open-weight model can match the expensive frontier LLM at roughly half the token cost. He argues businesses will adopt this for the cost savings, which would double productivity and margins. He cites the DeepSeek moment as an early sign and expects further cost cuts to follow.

  “I think a lot of that's going to shift to open weight... For 99% of the task, it's going to say, for 50% of what you're paying for a token, we can farm it to the best open weight model to perform the same task with 99% efficacy.”at 40:24

US–China AI & chip race

• Thomas J. Hayes · Great Hill Capital

  Blocking cheap Asian memory and chip imports would make the US lose the AI race to China.

  Hayes responds to the argument that the administration will never let US companies use Japanese, Korean or Chinese memory and chips. He says that if US firms cannot get enough domestic supply and cannot buy from lower-cost producers in China, Japan and Korea, the US will fall behind. His logic is that rejecting those exports means accepting a loss in the AI race, which he considers unlikely.

  “The administration will never let US companies use Japanese, Korean, and Chinese memory or semiconductors. Okay. Then you'll lose the AI race because they're all gonna be using it.”at 42:02