Oil Is the Headline. The Real Yield Is the Bill.

On September 27, Michael Howell of Capital Wars wrote that history shows the real economy turning fragile as long-term rates approach 5.5%, so the next large move in Treasury yields may be down. The next day Michael Burry wrote that rates are rising on inflation and on new debt competing with Treasuries, hyperscalerOne of the largest cloud and platform companies that build data centers at continental scale. More → borrowing among it, and that higher rates can make the chain financing the AI buildout come apart. The two readings contradict each other directly. Howell reads a 10-year yield above 5% as a growth cycle near its brake, and Burry as a cost the AI buildout's own borrowing helps raise.

The 10-year Treasury yield closed at 5.31% on October 5, its highest close since May 2002, and at 5.22% on October 8, by the Federal Reserve's daily series on FRED. It first closed at 5% or more on September 15; the next day the Fed raised its target range to 3.75% to 4.00%. Retail diesel reached a record $6.529 a gallon in the week of September 21, 71% above the $3.809 of the week of February 23, the Energy Information Administration's survey shows. On October 6 the agency's Short-Term Energy Outlook raised its forecast for Brent crude in the fourth quarter of 2026 to an average of $105 a barrel, $14 above the month before. Part fifteen asks whether the AI trade can survive a 10-year Treasury above 5% and an oil shock at the same time.

DeepStack read 167 posts, fund letters and episodes by 59 voices, published from July 15 to October 11, 2026, and verified 152 agreements and 171 disagreements, 64 of them direct contradictions and five refutations. Part seven asked who sets the long end, Part ten whether stocks are cheap against it and Part thirteen who pays the power bill. This Part asks by which channel the barrel reaches the AI multiple.

The boom pays the rate

The first reading holds that a 10-year above 5% is what growth costs, and that the boom can pay it. Howell argued on August 23 that long yields are rising because nominal growth has accelerated on a federal deficit near 6% of GDP and a large wave of AI capital spendingCash a company spends on long-lived assets: data centers, chips, networks, power. More →. On September 11 he counted term premiaThe extra yield investors demand to hold a long bond instead of rolling short ones. More → up 29 basis pointsOne hundredth of a percentage point: 25 basis points is 0.25%. More → since late June, against 51 for expected policy rates. Kris of Potential Multibaggers did the arithmetic on September 21: rate hikes do no harm while the economy and profits outgrow the rate, and an estimated 5.1% growth in third-quarter GDP plus inflation puts nominal growth above 8%.

Matthew C. Klein of The Overshoot wrote on September 1 that long yields are still low against reasonable expectations for inflation and growth. Ren found the side's evidence in the shares on October 5: the multiple on semiconductor stocks kept shrinking while their earnings kept growing, because the worry came from oil, yields, the Fed and the war.

Torsten Slok of Apollo answered on September 26 that more than a strong economy is lifting rates, with sticky inflation at the front end, record hyperscaler debt sales in the belly and fiscal worry at the long end. DeepStack found his account and Howell's count in direct contradiction. Howell carries his own limit. On September 18 he wrote that with higher oil, rising yields could begin to bite into growth, with the cycle perhaps 50 basis points from a major peak in the 10-year.

The two shocks compound

The second reading says oil and the long yield feed each other until equities give way. TSCS wrote on September 16 that both keep rising until stocks fall, because the S&P 500 is the only release valve the Treasury's buybacks leave open. Steve Eisman wrote on October 9 that only oil and the 10-year matter to the market day to day. He tied the 10-year's climb during the quarter, from 4.5% to over 5.25%, partly to about $500 billion of AI debt issuance crowding out Treasuries and partly to the Fed's September hike.

The inflation runs through core prices. Slok argued on September 23 that diesel sits in the cost of nearly every physical good, so its rise seeps after a delay into core goods and services, a pass-through the Fed cannot wave away as temporary. Krishna Guha, a guest on Eisman's program on October 4, said the Fed can no longer ignore the shocks after more than five years of inflation above its 2% target, dearer oil among the causes, and that with the job market in good shape a hike is defensible.

The borrowing runs through the same bonds. Slok explained on September 12 why it lands at the long end: data centers, power plants, transmission lines and federal deficits all draw on savings for decades. Mia Silverio of Prof G Markets named the cost on October 5: an investor who can earn 5.3% on a Treasury demands more from a stock and pays less for each dollar of its earnings. Daniel Romero of HyperTech Invest recounted on September 21 a stretch in which oil rallied 35% while semiconductor shares fell 30%. On this side, the AI trade is its own counterparty.

Oil stays out of the core

The third reading accepts the shock and doubts the cure. DeepValue Capital wrote on October 1 that rate hikes fight demand-led inflation while today's is supply-led, so no amount of tightening reopens the Strait of Hormuz. DeepStack judged that case a refutation of Guha's. Kris, in the same September 21 post as his growth arithmetic, called pricier oil the likely main reason for higher inflation and wrote that a higher rate will not cure it.

Kira Thornton of Variant Perception wrote on September 14 that core inflation trends remain muted, with Iran and El Niño leaving the energy and food tails beyond policymakers' reach. Howell crosses to this side on oil. On October 9 he wrote that diesel at current prices implies crude above $160 a barrel, short of a recession signal and consistent with a marked slowdown in real growth. DeepStack judged Slok's September 23 pass-through argument a refutation of Thornton's reading: diesel does not stay in the energy line.

Klein supplies the objection. In a September 14 note he wrote that producer prices for core goods and services still rise one to two percentage points faster than in the years before the pandemic, and that data from before the war with Iran contradict the claim that the war caused the excess. DeepStack judged the producer-price reading a refutation of DeepValue's September 18 case that hikes cannot cure an inflation born in the Strait. If the excess predates the barrel, the Fed is answering more than Hormuz, and a ceasefire would not end the hikes.

The bill comes as a real yield

TSCS measured the split on September 21: from the February 27 low the 10-year had risen 104 basis points, 96 of them real yieldThe yield on a Treasury whose principal rises with inflation: the price of money after inflation. More → and 8 breakeven inflation. Oil explains the daily wiggles in yields but not the trend, the post found: Brent was cheaper than at its spring peak while the 10-year stood 70 basis points higher, though since the war each $10 move in Brent had come with about 5 basis points on the 10-year. Romero wrote the same day that oil, consumer prices and yields move together, so rising oil forces the Fed's hand. DeepStack judged TSCS's split a qualification of Romero's link: the oil-to-yield channel runs through the real yield, not through inflation expectations. DeepStack re-ran the split on FRED on October 12, 2026.

From February 27 to October 8, 2026, the 10-year rose 125 basis points, from 3.97% to 5.22%. The 10-year real yield, read from Treasury inflation-protected securities, rose 115, from 1.72% to 2.87%, and closed at 2.95% on October 5, its highest since November 2008. Breakeven inflation, the market's ten-year inflation guess, rose 10, from 2.25% to 2.35%. The 5-year, 5-year forward rate, a gauge of the inflation expected five to ten years out, rose 22 to 2.32% on October 9. TSCS set the bar for the breakeven on September 16: cheap against realized inflation, and no bargain against its own two-year range of 2.18% to 2.46% at month-ends. At 2.35% on October 8 it sits inside that range, which narrows Slok's reading of record diesel crack spreadsThe extra yield a borrower pays above a Treasury of the same maturity, in basis points. More → and rising yields as a bond market pricing a broader inflation shock.

Howell wrote on September 18 that the rise came mainly from expected short rates, with term premia arguably falling; Hugh Vuillier of Variant Perception had written on September 6 that the real term premium drove it. The two contradict each other. The Fed staff's Kim-Wright model splits the difference by date: its term premium rose 26 basis points from June 30 to September 11, close to Howell's 29, and 62 from February 27 to October 2, nearly half of the 10-year's 131-point rise over those dates. TSCS cautioned on September 21 that the Fed's three models disagree about this year's leg. Michael W. Green of Yes, I Give a Fig wrote on August 23 that the structural buyers of long Treasuries have been withdrawing in stages for years, from Japan's hedged investors to corporate pensions. In the release-valve post of September 16, TSCS added that the buyers who once closed the gap, Japanese life insurers and U.S. insurers, have arithmetic reasons not to. The premium is what investors charge to hold durationHow far in the future a payment or a return sits, and how sensitive its value is to interest rates. More → while the Treasury and the hyperscalers sell it and, by Green's account, the old buyers step away.

DeepStack's reading is that the barrel helped move the Fed, landing on core prices that ran hot before the war, and that the Fed and the borrowers moved the real yield. Slok wrote on October 6 that a deal lowering oil would ease the pressure on rates, with 84% of global fixed income yielding more than 3%. If oil falls while the issuance continues, the real yield can hold, and a trade valued on earnings years away pays that rate. Howell's August 14 projection puts a 5.5% yield beside a market multiple below 20 times, about 10% lower, and Vitaliy Katsenelson named the likely casualties on September 10: expensive stocks whose cash flows lie far in the future.

Burry's warning concerns the lenders, and the first reading's own author shares the worry. Howell wrote on August 26 that if bond yields keep rising, the business of refinancing debt will groan under the strain, and that the creaking he hears now is literal. Part eight followed the loans behind the buildout. DeepStack's answer is that the damage so far shows in the price of the shares, in the shrinking multiple Ren found on October 5. The financing has held: by Slok's count on September 26, hyperscaler debt sales were running at a record. That answer fails if a hyperscaler cuts its capital-spending guidance or one of its long bond sales draws weak demand, and Alphabet and Microsoft report on October 28.

The tests ahead

The first result arrives on October 28. The Fed's decision, on its calendar, settles the test Part seven placed on the Docket, where DeepStack put the chance of a hike at 40 in 100. Alphabet reports that afternoon and Microsoft after the close. Third-quarter GDP and September's consumption prices follow on October 29, on the BEA's schedule, a check on Kris's nominal growth above 8%. October's consumer prices arrive on November 10, on the Bureau of Labor Statistics schedule, with the EIA's next outlook. Core goods would show Slok's diesel first.

DeepStack sets four dated tests, each read from a public source. Does the 10-year close at 5.50% or higher on any business day from October 16 through December 31, 2026? Does the EIA's outlook of January 12, 2027 put the fourth quarter's Brent average above the $105 it forecast on October 6, 2026? Does the 5-year, 5-year forward rate average above 2.60% in March 2027? Does the C Fund of the Thrift Savings Plan, the federal employees' S&P 500 index fund, close 10% or more below its highest close since October 16, 2026, on any business day through March 31, 2027, by its share price history?

DeepStack puts 35 on a 10-year close at 5.50% or higher by December 31, 2026, because Howell's brake sits there and Slok warned on August 29 that long rates six months on could be far lower. It puts 60 on Brent finishing the fourth quarter above the agency's $105, since the October 6 outlook raised that forecast $14 in one month and still describes extreme tightness in diesel. It puts 15 on the forward rate averaging above 2.60% in March 2027, and 40 on a 10% fall in the C Fund by March 31, 2027.

Two results would move the house to the second reading: a 10-year close at 5.50% or higher with breakeven inflation under its May 4 high of 2.50%, which would put the real yield past Howell's brake, or a forward rate above 2.60%. A Fed that holds on October 28 and a real yield back under 2.50% by the December 9 decision would mean the boom is paying the rate.

The AI trade is already paying for both shocks through one rate. The barrel gave the Fed one more reason to hike. The trade's own borrowing helped lift the real yield, and that is the yield its multiple answers to.