The week handed us three separate detonators that share the same fuse. August CPI printed 3.4% headline and a core that ticked one-tenth above consensus. Brent crude briefly cleared $108 on Thursday, its highest level since May, before Iranian state media floated Oman talks and pulled it back to roughly $104. And CME FedWatch finished the week pricing a 66% probability of a 25-basis-point hike at the September 16 FOMC meeting, a meeting that also produces an updated dot plot. Wednesday is not just about whether Warsh hikes. It is about what path he signals afterward, and that path is the thing that reprices every duration-sensitive name in the book.

The Dot Plot Is the Real Event

The market already has the hike substantially priced. CME FedWatch, tracking 30-day Fed Funds futures, put the probability of a quarter-point move at over 60% heading into the week. By Friday afternoon, after the CPI print, that number sat closer to 66%. A hike itself carries a "sell the rumor, buy the news" shape at this point, and any careful reader of equities, which bounced Friday with the S&P up 0.86% on the session, can see the market already partially digesting the decision. The far more consequential piece of paper lands alongside the decision: the Summary of Economic Projections.

The policy statement and press conference land September 16, and this is a Summary of Economic Projections meeting, meaning the Fed's updated dot plot will arrive alongside whatever decision it makes. The dot plot from June showed a modest tightening bias. If Wednesday's revision shows the median dot implying a second hike before year-end, every long-duration name in the book gets repriced instantly. That is not a hike thesis. That is a multiple-compression thesis, and it lands on top of an oil-driven inflation dynamic that the Fed cannot fully control.

The hike is nearly priced. The dot plot is not. Wednesday's real risk isn't 25 basis points, it's what the median Fed official now thinks comes next.

"Chair Warsh and others signaled that interest rates can remain on hold only if disinflation continues, and today's August report did not deliver that," said Kathy Bostjancic, Nationwide's chief economist, adding that "the renewed march higher in oil, gasoline and diesel prices add to concerns that higher energy prices could spill over to other goods and services." That is not a fringe read. That is the consensus interpretation of Friday's data, and it argues for a hawkish dot regardless of whether the committee moves Tuesday.

What Moved, and Why It Matters for the Pillars

The consumer price index rose a seasonally adjusted 0.4% for the month, putting the 12-month increase at 3.4%. That was in line with consensus, so the headline was not the story. The core CPI accelerated 0.3% for the month, a bit higher than expected, while the annual rate stood at 2.4%. One-tenth above on core is not a crisis. But it is a door-closer. The argument for holding September was always "disinflation continuing." The August CPI report is the last inflation update the Fed will see before next week's policy meeting, and it did not give the doves the cover they needed.

The energy story inside the number is the subplot worth watching. Gasoline prices surged 3.9% in August after falling 2.9% in July and were 27.4% higher than a year earlier, accounting for "over one third of the monthly all items increase." That surge is a direct transmission of the Iran/Hormuz risk premium into consumer prices. Large volumes of fertilizer also transit the Strait of Hormuz, threatening to push up global food prices. The oil shock is not yet finished printing through the CPI. September's report, due in October, absorbs a full month of Brent above $100.

For our book, this is a bifurcated read. COP and the energy sleeve are earning their keep as the explicit inflation hedge, and this week confirmed it. BAC and SCHW are more complicated: a flattening curve driven by a Fed-initiated hike squeezes bank lending spreads, but the brokerage model earns on client cash regardless of the curve shape, and higher-for-longer is categorically good for that float. The nuance bears repeating, because the market tends to sell the whole "financials" bucket when the curve compresses.

The price of oil surged Thursday, with Brent crude briefly surpassing $108 per barrel for the first time since May and WTI jumping above $102. Brent ended the session at $107.63, up 6.3% on the day, while WTI closed at $102.48, up 6.7%. The underlying driver was more structural than a single headline: Saudi Arabia informed OPEC that its crude oil output plunged in August to the lowest level since 1990 as a result of renewed hostilities with Iran.

The partial Friday reversal had a specific catalyst. Brent paused its rally to settle around $104 after Iranian state media said Tehran would meet Gulf states in Oman to discuss the Strait of Hormuz, with GCC diplomats expected to meet their Iranian counterpart Monday to discuss a possible temporary arrangement for managing shipping through the strait. "Possible temporary arrangement" is not a resolution. It is a ceasefire on the supply-shock narrative, and it is exactly the kind of headline that should be read with suspicion, because Monday's talks determine whether Friday's crude relief has legs or gives them back.

Analysts at ING said Thursday the move "reflects a market still pricing in persistent geopolitical risk, with Persian Gulf tensions showing no credible path to de-escalation." That framing matters for how we hold COP and interpret the power pillar. The energy-driven inflation dynamic does not resolve on Monday even if the diplomats shake hands. Saudi Arabia has shut its East-West crude pipeline as a precaution after multiple attacks, while Iran-backed Houthis reportedly advanced to Yemen's Perim Island. The supply disruption is already in the physical market, not just in futures.

Against a week that had every reason to punish growth, the AI hardware names did something notable: they mostly didn't break. Chipmakers gained on Friday, with AMD up 2.5% and Intel up 2.6%. Dell surged 11.9% to an all-time high on an RBC Outperform initiation. The Friday bounce in these names was partly short-covering against falling Treasury yields, but context matters: the fundamental numbers underneath the tape have not deteriorated.

NVDA's Q2 FY27 report delivered $96 billion in revenue with data center at $89 billion, and the company guided Q3 to $108 billion plus or minus 2%. Non-GAAP gross margin held at 75%. That is the floor beneath the valuation. Jensen Huang told the Street this is "a supply-constrained outlook" and said "we expect supply to remain a bottleneck at least through the end of fiscal year 28." A supply-constrained business has pricing power regardless of interest rates. That is the pillar-level logic that held through a week when the 10-year threatened 5%.

AMD posted a year-over-year jump of 50% in Q2 revenue and projects a 41% increase for the current quarter. The memory leg has its own moment coming: Micron reports its fiscal 2026 fourth-quarter results on September 30. We own NVDA, AMD, MU, and TSM across the AI chips pillar, and this week did not hand us a single thesis-level crack in any of them. Price volatility driven by rising real rates is not the same thing as an erosion of demand. We have said that before. The week's tape made the case again.

NVIDIA and Palantir announced they are bringing sovereign intelligence to critical supply chains (September 10). Read that for what it is: another incremental proof point that the government and enterprise AI deployment cycle is expanding beyond the hyperscalers into mission-critical infrastructure. That is the next leg of NVDA's revenue story, and the export ban on China datacenter is already excluded from the guidance numbers, meaning these new sovereign channels are pure upside to the model.

This is the least comfortable section to write, but the honest one. The 10-year Treasury yield rose as high as 4.95% on Thursday, its highest since 2023, as Treasuries continued to sell off. Real short rates remain solidly positive: 3-month bills at 3.91% against a 3.4% CPI print. The repression thesis holds that the government will manage rates lower to handle the debt stock. The tape this week was arguing the opposite. That tension is the defining macro question in the book right now.

Bitcoin failed the $78,000 shelf again this week, settling Friday near $77,244. Gold closed around $4,390, off its recent highs. These are not catastrophic moves, but the direction is clear: when real rates are positive and widening, the assets that depend on negative or suppressed real yields stop working as they were designed to. We hold GLD and IBIT as the hard-asset sleeve and the live test of the repression frame. The FOMC decision and, more importantly, the dot plot will either validate or challenge that thesis in a single afternoon. If Wednesday's dots show a Fed committed to running real rates positive through 2027, the repression frame needs a hard re-examination.

What to Watch Next Week

Wednesday, September 16 is the only date that matters. After being on hold throughout 2026, the Fed is expected to approve a quarter-percentage-point hike of its benchmark interest rate, though market pricing has been volatile. Public statements from Fed officials have been split, with Chairman Kevin Warsh recently emphasizing his commitment to getting inflation back to target and saying action may be needed. The precise language around "data-dependent" versus a signaled path will determine whether rate-sensitive names stabilize or reprice again next week. A hike with a dovish tilt in the dots lets the book breathe. A hike with a second move sketched for November turns the pressure back on.

Monday's GCC-Iran talks in Oman are the geopolitical tripwire. GCC diplomats are expected to meet their Iranian counterpart Monday to discuss a possible temporary arrangement for managing shipping through the strait. The two-word phrase "temporary arrangement" does all the analytical work. No structural Hormuz resolution is priced in; even partial progress on commercial shipping lanes would compress the oil risk premium and provide meaningful relief for the energy-driven portion of the CPI. A breakdown in talks reverses Friday's crude giveback instantly.

The Bank of Japan meets September 17-18, a session that warrants watching even if the primary market is domestic. The yen has strengthened roughly 4% over three weeks while U.S. yields rose simultaneously, which is the signature pattern of foreign selling of Treasuries rather than a flight to safety. If the BOJ signals any further normalization, the yen-carry unwind pressure on U.S. long rates becomes a secondary source of upward yield movement, entirely independent of the Fed's own decision. Two central banks, same direction, same week. That is a rate-volatility setup, not a calm one.

Micron reports its fiscal 2026 fourth-quarter results September 30, which is not next week but is close enough that the setup conversation is already happening. The MU trade is not just memory, it is the HBM (high-bandwidth memory) story that sits directly underneath the NVDA data center buildout. If MU's HBM order visibility confirms what NVDA's guide implies, the memory cycle thesis gets a third consecutive confirmation. That matters for the chips pillar's conviction heading into Q4.

Forward Thesis by Pillar, Conviction and What Would Change It

AI Chips / Hardware (NVDA, AMD, MU, TSM): Conviction holds. The thesis is supply-constrained compute paid by hyperscalers under multi-year contracts, and nothing that happened this week challenged the demand side. NVDA's $108 billion Q3 guide, AMD's 41% projected revenue growth next quarter, and the Palantir sovereign AI partnership all add evidence. Rate pressure from the front end affects the multiple, not the revenue. The risk to watch is a sustained 10-year above 5.00% that compresses growth multiples across the board, but that is a valuation conversation, not a thesis conversation, and we know the difference.

Power and Grid (VST, CEG): The pillar is repairing rather than leading. A rising discount rate hits power infrastructure twice: once through higher financing costs for capital-intensive assets, once through the discount rate applied to long-duration contracted cash flows. The fundamental demand story, 24/7 firm power for datacenter PPAs, is intact and arguably strengthening as AI infrastructure spending accelerates. But "right thesis, wrong moment" is a real category, and this is it. We are watching the pillar, not abandoning it.

Platforms and AI Software (MSFT, AMZN, NOW): AWS remains the cleanest expression of hyperscaler capex turning into enterprise revenue, with the reacceleration thesis intact. MSFT is the capex owner on the other side of the chip trade. NOW remains on the shortest leash in the book, because enterprise AI monetization can be deferred when CFOs get nervous about macro. A hike plus hawkish dots creates exactly the macro environment in which software evaluations slow. That is not a thesis exit for the platform names, but it is a reason to hold the conviction measured against the actual evidence in the next earnings cycle.

Financials / Rate-Regime (BAC, SCHW): This is the pillar that actually benefits from the week's outcome, with important nuance. A Fed-driven curve flattening squeezes bank net interest margin, which is a headwind for BAC's lending book. But SCHW earns on the float of client cash balances, and that float yield rises with the policy rate regardless of curve shape. Higher-for-longer is structurally favorable to the brokerage model in a way it is not always favorable to the commercial bank. We hold both names; the read on each is different this week.

Hard Assets (GLD, IBIT): The honest assessment: this sleeve is on the wrong side of the current rate regime. Positive real rates are the primary adversary of assets that carry no yield, and 3-month bills at 3.91% against 3.4% CPI is a real rate of roughly +50 basis points. Bitcoin failing the $78,000 shelf twice in the same week is a tape observation, not a panic signal, but it is consistent with the regime. The repression thesis holds as the structural frame, the debt stock remains too large to service at real rates indefinitely, but the timeline for when repression reasserts itself is uncertain, and the intermediate path runs through Wednesday's dot plot. Wednesday matters.

Energy (COP): The most straightforward pillar in the book this week. Crude up nearly 10% on the week, Brent touching $108, Saudi output at three-decade lows, and an ongoing Hormuz risk premium embedded in the curve, every one of those facts is a direct confirmation of the thesis. COP's Permian volumes are priced in dollars-per-barrel; when that barrel is worth $100 rather than $80, the free cash flow math changes materially. We are watching Monday's Oman talks closely because a credible diplomatic de-escalation scenario would compress the risk premium and partially reverse this week's gains. That is a risk to manage, not a reason to abandon the position.

Off-AI (VRTX) and Frontier (TSLA): VRTX remains the deliberate uncorrelated position in the book, with CF cash flows and a genuine pipeline in non-opioid pain and Type 1 diabetes that does not care what the 10-year does. No news this week that changes that read. TSLA is long-horizon autonomy and physical AI optionality, sized accordingly, and the less said about it in a week defined by macro, the better.

The frame itself: The repression thesis is being tested by its own falsifier this September. A Fed that hikes and then signals more, over a base of energy-driven inflation that it cannot directly control, is the scenario in which real rates stay positive and the hard-asset sleeve continues to underperform. The competing scenario, repression eventually wins because the debt stock leaves no other option, is the structural thesis, and it remains intact as a multi-year frame. The market is asking how long the interim pain lasts. Wednesday gives us the next data point.

Nova One Capital and the Nova X Convergence Fund hold positions in securities discussed in this letter, including NVDA, AMD, MU, TSM, MSFT, AMZN, NOW, BAC, SCHW, GLD, IBIT, COP, VRTX, VST, CEG, and TSLA. This letter is for informational purposes only and does not constitute investment advice, a solicitation, or an offer to buy or sell any security. Past performance is not indicative of future results. All views represent the opinions of Nova One Capital as of September 12, 2026, and are subject to change without notice. Investors should conduct their own due diligence before making any investment decisions.