This was a week that asked two enormous questions and got one enormous answer. The answer came Wednesday night from Santa Clara: $96.2 billion in quarterly revenue, growth of 106% year over year, and guidance for roughly 70% growth in fiscal 2028, nearly twice what the Street had penciled in. The question that didn't get answered came Friday morning from a lodge in Wyoming, where Federal Reserve Chair Kevin Warsh delivered his inaugural Jackson Hole keynote and managed to say almost nothing actionable about rates, inflation, or the path forward. The market's response to this asymmetry was telling: the S&P 500 closed Friday at 7,711, the Nasdaq at 26,402, and the VIX settled at a remarkably placid 14.43. Investors decided, correctly, that a blowout NVDA quarter was worth more than a cryptic Fed chair, and bid the week higher on that logic.

The Lead: $96 Billion Is Not a Bubble

Nvidia reported earnings of $2.22 per share, more than double one year ago, on revenue of $96.2 billion, up 106% year over year. Those are the numbers, but they are not the story. The story is in the details that followed.

The company maintained its gross margin at 75%, though it expects that to dip next quarter due to rising memory costs. The real surprise was guidance for revenue to grow even faster next year, with roughly 70% growth guided for calendar 2027, nearly twice what analysts had expected. When a company of this size guides for acceleration, not deceleration, the "demand is peaking" thesis doesn't just lose, it gets publicly embarrassed.

Amazon and Nvidia announced that AWS will buy 2 million Nvidia GPUs and deploy the company's new Vera CPU, with some integrated into Nvidia's forthcoming Rubin chip. The deal confirms that hyperscalers continue to spend heavily despite investor fears that they may be tapping out. That fear has been the central bear argument on this pillar for six months. It is now harder to maintain with a straight face.

CFO Colette Kress said capex among the "top five hyperscalers" is expected to increase to $1.3 trillion next year from $800 billion in 2026. Read that number twice. The spend is not flattening. The spend is going vertical. And Nvidia, sitting at the center of that flow with a 75% gross margin, is collecting a toll on every dollar of it.

Jensen Huang said AI "reached its inflection point," noting that the number of companies that need large clusters of GPUs has expanded dramatically. "This time last year, one lab alone was driving the buildout," Huang said. That observation is the most important sentence in the earnings release. The demand base has broadened from a handful of frontier labs to an entire tier of enterprise and sovereign buyers. That is not a story that ends next quarter.

Nvidia's supply commitments more than doubled to $279 billion, a figure that tells you everything about the forward visibility embedded in this business. We own NVDA. This quarter is not a reason to revisit that. It is a reason to be comfortable with it.

The AI compute buildout didn't peak. It changed shape, broadened its customer base, and accelerated its forward guidance. $96 billion in one quarter is not a bubble. It is a new baseline.

The Hedge: Warsh Came, Spoke, and Left No Clue

The Federal Reserve Bank of Kansas City convened its annual Economic Policy Symposium August 27-29 in Jackson Hole, Wyoming, with this year's theme exploring "Financial Innovation: Implications for Payments and Policy." The theme was, for most market participants, beside the point. What mattered was what Warsh would say about inflation, rates, and the September 16 FOMC meeting.

Warsh expressed concern about elevated inflation while hinting that interest rates could need to move higher if more progress isn't made on easing price pressures. His closely watched remarks avoided committing either to forward guidance or to a reaction function, the economic signals that would warrant an adjustment in rates. That last part is the key. A Fed chair without a publicly stated reaction function is a Fed chair who retains maximum optionality, which is another way of saying the bond market has to do its own guessing.

Warsh's speech differed from his predecessors, who often utilized Jackson Hole either as a cue on the direction of rates, broad changes to the Fed's framework, or to delineate new ways of approaching monetary policy. At last year's event, then-Chair Jerome Powell hinted at rate cuts ahead, setting off an aggressive rally. Warsh was the anti-Powell on Friday. Where Powell used Jackson Hole as a signal, Warsh used it as a seminar.

Treasury yields at the short end of the curve spiked while those on the long end edged lower after Warsh said the central bank still has "work to do" to bring inflation under control. The 2-year Treasury yield jumped more than 6 basis points to 4.298%, while the 30-year bond yield was 2 basis points lower at 5.168%, and the 10-year was little changed at 4.676%. That yield curve response is the market's actual verdict: the short end repriced hawkish optionality, but the long end caught a bid, suggesting investors read the speech as slightly reassuring on the tail risk of runaway long yields.

Context matters here. The 30-year Treasury yield had closed at 5.31% on August 17, its highest since 2007, and the Treasury Department had stepped into the bond market on August 19 to try to bring long-term borrowing costs down. From that starting point, a Friday close near 5.17% on the long bond is measurable relief, even if it came more from Treasury's buyback intervention than from anything Warsh actually said.

Inflation remains stubbornly above the Fed's 2% target at 3.4%, while economic growth shows signs of cracking under the weight of trade tensions and geopolitical uncertainty. Roughly half the FOMC penciled in rate hikes for 2026 at Warsh's first meeting in June, with three regional Fed presidents dissenting in favor of immediate tightening at the July meeting. That is not a unified committee. A divided FOMC behind a chairman who won't reveal his reaction function is, to put it simply, a rate regime that the bond market has to price by feel. For long-duration growth assets, that is the real lingering overhang. The NVDA print can overwhelm it for a week. It cannot make it disappear.

We own BAC partly for this reason. A bank that earns more money as short rates stay elevated and the curve stays steep doesn't need the Fed to cut. It needs the Fed to stay confused, which is exactly what it appears to be doing.

The Tape: What Else Moved and Why It Matters

The geopolitical texture of the week was meaningfully less alarming than the prior two. Crude oil prices closed more than 3% lower Tuesday as the U.S. shifts to a strategy of economic pressure on Iran, easing market fears of a renewed war in the Persian Gulf. Treasury Secretary Scott Bessent vowed "the greatest coordinated economic isolation in the history of the world" against Iran, adding that if maximum economic pressure is applied, "there will not be a large-scale kinetic restart." The Hormuz fear trade that had been lifting oil and compressing equity multiples for weeks finally got a pressure valve. Brent retreating off its highs is a quiet tailwind for both the consumer and the rate picture.

That said, the trade front produced a new wrinkle. On Tuesday, Canada said it would slap retaliatory tariffs against the United States, applying to roughly $20 billion in goods. The new tariffs, set to take effect on September 8, would apply to more than 700 U.S. goods and range from 15% to 50%, targeting steel, aluminum, dairy, seafood, and more. Ottawa's new levies came after the U.S. announced additional 50% tariffs on a variety of Canadian products, targeting motor vehicles, alcohol, and dairy. The Canada escalation is worth watching for its second-order effects on input costs, particularly in steel and aluminum, which flow directly into the grid and power infrastructure buildout that underpins our power pillar. So far it is noise. If September 8 arrives without a negotiated pause, it becomes something to price.

MP Materials continues to hold its own as the reshoring thesis firms. The rare-earth story is not a function of any single week's price action; it is a function of the export control regime and the DoD supply chain policy that gives MP a contracted revenue floor independent of spot NdPr prices. We are long MP. The thesis is intact and the macro environment, featuring a U.S. administration pursuing maximum economic separation from China in both energy and critical minerals, continues to validate it.

On the security software side, PANW has been a strong performer through August and the earnings event that matters most for that pillar is now two days away. Palo Alto Networks is confirmed to report fiscal year-end results on September 1, 2026, after the close. Cyber cohort earnings this season have punished extended names even on beats, which is the relevant setup risk. The AI-native security platform thesis at PANW is solid. The question, as always, is what the market already has priced.

LLY continues to do what a diversifier is supposed to do: not correlate. The GLP-1 market structure thesis, a runway measured in years and patient populations measured in the hundreds of millions, doesn't change on a week where the bond market is repricing Warsh risk. We own it for exactly that reason.

Gold, held through GLD, continues to earn its strategic allocation. The Warsh speech and its explicit non-commitment on the rate path, a weaker dollar on balance, and central bank buying that has been structural for two years are all constructive for the thesis. This is not a panic trade. It is a debasement hedge that the market keeps finding new reasons to validate.

The Setup: What's on Deck

The September 4 jobs report is the week's fulcrum. Warsh delivers his first keynote just three weeks before the critical September FOMC meeting, with markets pricing in roughly one-in-three odds of a September rate hike, making monetary policy direction a live question. That uncertainty doesn't resolve until payrolls print. A soft number (below 100,000) would reopen the September cut conversation that Warsh just sealed shut. A strong number (above 175,000) would validate the "inflation work to do" framing and put a September hike back on the table in a serious way. The market is priced for the middle of that range, which means the tails are where the action lives.

PANW on Monday night is the pillar-specific event. The platform transition that began 18 months ago, consolidating point products into an AI-native security platform, was the right strategic call. The question for fiscal Q4 and the FY27 guide is whether the Annual Recurring Revenue growth, particularly in Next-Generation Security ARR, has reached the velocity that justifies the valuation. This season has been unforgiving to anything that disappoints on forward ARR guidance, and we will be watching that number more closely than the headline EPS beat.

The Read: Forward Thesis by Pillar

AI Chips / Hardware (NVDA, AMD, MU). NVDA's fiscal Q2 report is the strongest single-quarter confirmation the AI infrastructure thesis has received. Revenue doubled, margins held, supply commitments ballooned to $279 billion, and the demand base broadened from a handful of labs to an ecosystem of enterprise and sovereign buyers. The thesis is not that NVDA will grow this fast forever. The thesis is that the installed base is supply-constrained, contracted, and paid in advance, making the revenue stream far more durable than a cyclical hardware story. What would change it: a meaningful hyperscaler capex revision downward in Q3 guidance calls, or a gross margin collapse below 70% as memory costs bite. Neither appears imminent.

Power & Grid (GEV, VST, CEG). Firm power and the wires remain the physical bottleneck of the AI buildout, and the hyperscaler capex trajectory guided this week, from $800 billion in 2026 to $1.3 trillion in 2027, makes that bottleneck tighter, not looser. The thesis is straightforward: you cannot run a $1.3 trillion AI infrastructure buildout on intermittent power. You need firm megawatts and the transmission lines to carry them. What would change it: a structural change in hyperscaler demand (contradicted by this week's data), or a regulatory delay in grid interconnection timelines.

AI Software / Security (PANW, NOW). PANW reporting Monday is the near-term catalyst. The conviction here is that the consolidation of enterprise security onto AI-native platforms is a multiyear structural trend, not a product cycle. Next-Generation Security ARR growth is the metric that matters. What would change it: a forward ARR guide that implies decelerating platform adoption, or a macro-driven IT spending freeze that compresses the deal pipeline.

Critical Minerals / Reshoring (MP Materials). The thesis is unchanged and the macro environment, maximum economic pressure on China in both trade and technology, is its most favorable backdrop since the company went public. The DoD price floor and Apple offtake underwrite a minimum earnings stream that the market continues to underprice relative to the infrastructure-like nature of those contracts. The near-term risk is a US-China diplomatic breakthrough on export controls that narrows the ex-China rare-earth premium. We watch that tail carefully. We are long MP.

Fuel Cycle / Energy Security (LEU). A repairing, conviction-level hold. The HALEU enrichment reshoring story is a national security mandate with a DOE award behind it. It does not trade on weekly headlines. It trades on domestic enrichment capacity timelines and the long-dated energy security agenda, both of which remain intact.

GLP-1 / Healthcare (LLY). The strongest non-AI diversifier in the book. The metabolic drug market is a decade-long revenue opportunity and LLY is its most defensible incumbent. The short-term rate uncertainty that pressures long-duration growth assets applies here too, but the earnings visibility and the market size make LLY a willing-to-hold-through-volatility name. What would change it: a safety signal that widens the label restrictions, or a biosimilar approval timeline that accelerates meaningfully.

Financials / Rate Regime (BAC). A Warsh who refuses to guide and a long end that stays elevated is, counterintuitively, a constructive backdrop for a bank with NII momentum and a clean balance sheet. BAC earns more when short rates stay high and the curve stays steep. The September FOMC meeting is the event to watch; we will be listening to the dots for any shift in the 2026-2027 path. What would change the thesis: a rapid capitulation to cuts that compresses the net interest margin faster than the loan growth can offset it.

Gold (GLD). The Warsh framework, opaque on the reaction function and explicit about inflation work remaining, is structurally bullish for gold. A central bank chair who won't commit to a rate path is a chair who leaves the inflation tail open, and gold prices that tail. The debasement bid from global central bank buying is structural, not tactical. We hold GLD as a strategic allocation that earned its place on its own signal.

Nova One Capital and the Nova X Convergence Fund hold positions in securities discussed in this letter, including NVDA, GEV, VST, CEG, PANW, NOW, MP Materials, LEU, LLY, BAC, and GLD. This letter reflects the firm's standing positioning and thesis as of August 29, 2026. This document 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 investing involves risk, including the possible loss of principal. Nothing herein should be construed as a current or intended trade signal. Published August 29, 2026.