NVIDIA reports after close today — the single most important earnings in the market. Trump went 1 hour 47 minutes last night and said almost nothing about Iran. Gold hit $5,183. VIX still elevated at 19.5 after Monday's spike to 21. The tariff rate is still legally indeterminate. AMD is still ripping on the Meta deal.
Crypto data provided by CoinGecko
NVIDIA reports after the close today — this is the event. $65.7B revenue consensus (+67% YoY), $1.53 EPS (+72% YoY). Polymarket: 94.5% probability of a beat. Goldman modeling $67B+. But the headline beat is priced — what matters is the texture underneath. Blackwell margin trajectory tells you whether the H100→Blackwell transition is margin-dilutive. China revenue tells you how much the H200 permission (with 25% US revenue share) is actually worth. Q1 FY27 guidance (Street: ~$71B) tells you if the demand curve is still accelerating or flattening. Management's goal is mid-70s gross margins by year-end — if they're not tracking toward that, the market will care more about the margin miss than the revenue beat. The entire semiconductor complex is positioned around tonight's print.
Trump's State of the Union ran 1 hour 47 minutes — a record. The substance was thin relative to the length. Defended his economic record with "a turnaround for the ages." Reiterated the institutional investor housing ban — potentially meaningful for residential real estate if enforced. Honored the US hockey team. On Iran: "I would never let Iran have a nuclear weapon" but offered zero specifics on the military buildup, the Geneva talks Thursday, or Araghchi's claim that the US hasn't demanded zero enrichment. On Ukraine: "a war which would have never happened if I were president" — said on the four-year anniversary of the invasion with no update on aid or diplomacy. On tariffs: defended the approach without addressing the SCOTUS loss or the 10% vs 15% confusion. Virginia Gov. Spanberger delivered the Democratic response. 60% disapproval going in. 15+ Democrats boycotted. The market priced the speech as a non-event, which is what it was.
The tariff regime is in legal limbo and getting worse. The 10% Section 122 tariff took effect Monday. Trump announced 15% but the formal order is still pending — customs is implementing 10%. Meanwhile, the SCOTUS 6-3 ruling struck down all IEEPA tariffs, and the majority deliberately left the $160B refund question open. The Section 122 replacement is itself legally vulnerable — Trump's own DOJ previously said the statute doesn't apply. Former IMF Deputy Gita Gopinath publicly rejected the "balance of payments crisis" justification. EU froze trade deal ratification. Japan seeking assurances on its $550B investment pledge. India-US trade talks on hold. South Korea reviewing. The tariff rate is indeterminate, the legal basis is contested, and every major trade partner is recalibrating simultaneously. For businesses, this is worse than a high rate with certainty.
DXY at 97.66 — down 8% over the past year, three consecutive months of decline. Fed's Collins: holding rates steady "likely appropriate." Barkin: "well-positioned." 98% probability of unchanged in March. But markets are pricing three cuts this year. The disconnect between official caution and market expectations creates a rate-sensitive setup for the March 6 jobs report and March 11 CPI. Dollar weakness supports gold and commodities but complicates the tariff calculus — a weaker dollar makes imports more expensive, which amplifies the tariff impact on consumer prices.
DHS shutdown Day 11 — the longest DHS-specific funding gap in history. Congress returned from recess yesterday. SOTU produced no deal. FEMA disaster relief at $9.6B (down from $30B). Global Entry suspended. TSA workers miss first full paycheck March 14 if unresolved. Mullin predicts resolution this week. The shutdown is now converging with the Iran deadline (Thursday), NVIDIA earnings (tonight), and tariff confusion into the densest week of policy risk in 2026.
BTC bounced to ~$65,500 after Monday's flush to $62,858 — relief, not reversal. The 4% bounce is happening on light volume against a backdrop of record-negative US demand (40 consecutive days per CryptoQuant). BTC is still 48% below ATH, deeply under the 50D (~$82K) and 200D (~$85K). The $60K psychological level remains the key support — if it breaks, the 200W MA at ~$42K becomes the conversation. ETF flow data later this week will tell us whether the institutional selling that accelerated last week has paused or was just a one-day reprieve.
Vitalik sold 17,000 ETH this month as ether fell 37%. Tracked wallets dropped from 241K ETH to 224K — a $32M drawdown in founder holdings during a bear market. ETH at $1,915 is bouncing but the selling from the project's own founder creates a persistent headwind. The bull case: whale accumulation on-chain below $1,900 suggests smart money is buying what Vitalik is selling. The bear case: the founder knows something about demand the market doesn't.
SOL range-bound at $82 — two weeks of consolidation in the $75-$85 zone. Strategic reserve inclusion buzz is providing a sentiment floor but the fundamental agent commerce thesis hasn't produced new data points. Solana is still leading all chains in payment volume during the bear, which matters for the structural case. The contrarian setup from last week (deeply negative funding rates + persistent ETF inflows) remains intact. If macro stabilizes, SOL has the most explosive upside of the three majors from these levels. If it doesn't, the $50 level is the next test.
Total crypto market cap at $2.33T — BTC dominance 56.1%, ETH 9.9%. The dominance spread tells the story: capital is concentrating in BTC during the downturn, which is textbook bear market behavior. ETH dominance below 10% is a multi-year low. The rotation back into alts happens when BTC stabilizes and the risk curve steepens — we're not there yet.
NVIDIA earnings tonight — the most consequential report of the quarter. Management guided for Blackwell revenue to jump from $7.1B last year to an implied $93.7B this year. China sales remain uncertain — the H200 permission came with a 25% revenue share to the US government. The market wants clarity on: (1) Blackwell margin ramp, (2) China order trajectory, (3) Q1 FY27 guide (Street expects ~$71B). If NVIDIA beats on all three, the S&P likely breaks above the 200D.
AMD +8.8% on Tuesday from the Meta deal. Stock at $213.84. The $60-100B deal with equity warrants (160M shares, ~10% of AMD) continues to digest. Lisa Su: "The CPU market is absolutely on fire." AMD now has a credible claim to be NVIDIA's peer in AI infrastructure, not just its alternative.
Spotify engineers stopped writing code. CEO Gustav Söderström says the most senior developers haven't written a line of code since December — they supervise AI-generated output instead. This is the quietest data point of the week and possibly the most important. When a $100B company's best engineers become reviewers instead of writers, the labor market implications are enormous.
AI spending projected at $2.5T in 2026 — 44% increase over 2025 per Gartner. Enterprise AI agent adoption going from 5% to 40% of apps by year-end. The infrastructure buildout is accelerating into the spending. Meta ($115-135B), Alphabet ($175-185B), Amazon ($200B). The Big Five hyperscalers have pledged over $650B in 2026 capex alone — nearly double 2025.
Iran-US nuclear talks resume Thursday in Geneva — third round. Iran ruling out an interim deal, demanding comprehensive sanctions relief. Araghchi's claim that the US "has not asked for zero enrichment" was not addressed in the SOTU. The military buildup remains the largest since 2003. Oil at $66 pricing cautious optimism but the risk premium is building ahead of Thursday.
SCOTUS IEEPA ruling aftermath continues to ripple. The $160B in tariffs collected under IEEPA — no clarity on refunds. The majority opinion deliberately left the remedial question open. The Section 122 replacement is legally vulnerable — Trump's own DOJ previously said the statute doesn't apply. If courts strike Section 122, the tariff regime collapses entirely.
DHS shutdown enters Day 11. Congress returned from recess yesterday. SOTU didn't produce a deal. TSA workers miss first full paycheck March 14 if unresolved. FEMA disaster relief at $9.6B (down from $30B). Global Entry still suspended.
$650 billion. That's what the five largest US technology companies have collectively pledged to spend on AI infrastructure in 2026 alone. Meta: $115-135B. Alphabet: $175-185B. Amazon: $200B. Microsoft and Oracle filling the gap. This is roughly the GDP of Switzerland — committed to data centers, GPUs, and cooling systems in a single calendar year.
NVIDIA's earnings tonight, after the close, are the first real-time audit of whether that spending is translating into returns.
The consensus numbers are almost beside the point. $65.7B revenue (+67% YoY), $1.53 EPS. Polymarket puts 94.5% odds on a beat. Goldman is modeling $67B+. Everyone expects the headline to be green. The question is what's beneath it.
Three things determine whether this is a catalyst or a ceiling. First, Blackwell margins. Management has guided toward mid-70% gross margins by year-end. If Blackwell is ramping but margins are compressing, it means the shift from high-margin H100 to lower-margin Blackwell is happening faster than the mix can absorb. That's a growth-quality downgrade even if the topline beats. Second, China. The recent permission to sell H200 chips with a 25% revenue share to the US government creates a new, murky revenue stream. How much is real? How much is modeled but unconfirmed? Third, Q1 FY27 guidance. The Street expects ~$71B. If NVIDIA guides above that, the AI capex flywheel narrative remains intact. If they guide in-line or below, it raises the question the AMD-Meta deal was supposed to answer: is the TAM expanding fast enough to absorb multiple suppliers?
The AMD-Meta deal reframed the competitive landscape this week. $60-100B in GPU commitments. 6 gigawatts. Equity warrants for 10% of AMD. When Meta needs both NVIDIA and AMD at scale, the bull case writes itself: demand exceeds supply, multiple winners. But the bear case also sharpens: if Meta is diversifying because NVIDIA's pricing power is peaking, then NVIDIA's margins compress from competition, not from execution failure. Tonight's margin data reveals which story is true.
What the market isn't pricing correctly is velocity. The hyperscalers collectively doubled their AI capex commitments from 2025 to 2026. Spotify's most senior engineers stopped writing code in December. Gartner says AI agent adoption in enterprise apps jumps from 5% to 40% this year. The demand side isn't growing linearly — it's compounding. If NVIDIA's results confirm that the supply side can keep pace, this market has significantly more room to run. If not, the GPU bottleneck becomes the new story, and the investment thesis shifts from "buy the picks and shovels" to "buy the scarcity."
The honest framework: NVIDIA will almost certainly beat. The question is whether the beat validates the $650B capex thesis or reveals the first cracks in it. Pay attention to three numbers: gross margin percent, China revenue dollars, and Q1 guidance. Everything else is noise.
"Between stimulus and response there is a space. In that space is our freedom and our power to choose our response. In our response lies our growth and our freedom."
— Viktor Frankl
You probably have a default reaction speed. Something happens — an email, a comment, a piece of news, a feeling — and you respond. Fast. Automatically. The speed feels like competence. It's usually just habit.
Frankl's insight, forged in circumstances incomparably harder than anything most of us face, is deceptively simple: there's a gap between what happens to you and what you do about it. Most people don't experience that gap because they've trained themselves to close it immediately. The response becomes the stimulus — an unbroken chain of reaction that feels like living but is actually just reflex.
The highest-leverage skill you can develop isn't a faster response. It's a longer pause. Not indecision — deliberate space. The kind of space where you actually choose instead of react. Every meaningful upgrade in how you think, relate, and create comes from widening that gap by even a few seconds.
Pick the three highest-stimulus moments of your day — the ones where you usually react fastest. Before each one, take one full breath and ask: "What am I about to do, and is that what I'd choose?" You don't have to change anything. Just notice the gap. That's where the growth lives.
# ▸ THE TAKE
NVIDIA Tonight Is a Referendum on the Biggest Bet in Business History
$650 billion. That's what the five largest US technology companies have collectively pledged to spend on AI infrastructure in 2026 alone. Meta: $115-135B. Alphabet: $175-185B. Amazon: $200B. Microsoft and Oracle filling the gap. This is roughly the GDP of Switzerland — committed to data centers, GPUs, and cooling systems in a single calendar year.
NVIDIA's earnings tonight, after the close, are the first real-time audit of whether that spending is translating into returns.
The consensus numbers are almost beside the point. $65.7B revenue (+67% YoY), $1.53 EPS. Polymarket puts 94.5% odds on a beat. Goldman is modeling $67B+. Everyone expects the headline to be green. The question is what's beneath it.
Three things determine whether this is a catalyst or a ceiling. First, Blackwell margins. Management has guided toward mid-70% gross margins by year-end. If Blackwell is ramping but margins are compressing, it means the shift from high-margin H100 to lower-margin Blackwell is happening faster than the mix can absorb. That's a growth-quality downgrade even if the topline beats. Second, China. The recent permission to sell H200 chips with a 25% revenue share to the US government creates a new, murky revenue stream. How much is real? How much is modeled but unconfirmed? Third, Q1 FY27 guidance. The Street expects ~$71B. If NVIDIA guides above that, the AI capex flywheel narrative remains intact. If they guide in-line or below, it raises the question the AMD-Meta deal was supposed to answer: is the TAM expanding fast enough to absorb multiple suppliers?
The AMD-Meta deal reframed the competitive landscape this week. $60-100B in GPU commitments. 6 gigawatts. Equity warrants for 10% of AMD. When Meta needs both NVIDIA and AMD at scale, the bull case writes itself: demand exceeds supply, multiple winners. But the bear case also sharpens: if Meta is diversifying because NVIDIA's pricing power is peaking, then NVIDIA's margins compress from competition, not from execution failure. Tonight's margin data reveals which story is true.
What the market isn't pricing correctly is velocity. The hyperscalers collectively doubled their AI capex commitments from 2025 to 2026. Spotify's most senior engineers stopped writing code in December. Gartner says AI agent adoption in enterprise apps jumps from 5% to 40% this year. The demand side isn't growing linearly — it's compounding. If NVIDIA's results confirm that the supply side can keep pace, this market has significantly more room to run. If not, the GPU bottleneck becomes the new story, and the investment thesis shifts from "buy the picks and shovels" to "buy the scarcity."
The honest framework: NVIDIA will almost certainly beat. The question is whether the beat validates the $650B capex thesis or reveals the first cracks in it. Pay attention to three numbers: gross margin percent, China revenue dollars, and Q1 guidance. Everything else is noise.
Information is measured by the number of alternatives it eliminates. It's the fundamental currency connecting energy, matter, and consciousness. The physical world isn't just colliding atoms—it's a network of correlations and information exchanges between systems. Information is finite, observer-dependent, and inseparable from physical reality.
Understand that information processing has physical costs. Quantum mechanics shows you can't learn for free—gaining information requires energy and increases entropy elsewhere. There's a thermodynamic cost to knowledge.
# ▸ THE BIG STORIES The macro trends that matter through the daily noise. Updated when news moves the needle. Silent when it doesn't.
Current state: Report due after close today. $65.7B revenue consensus, $1.53 EPS. Blackwell ramp, China sales, and Q1 guidance are the three variables that matter. Today's update: The setup is maximally crowded — 94.5% beat probability on Polymarket, 37 buy / 1 hold / 1 sell from analysts. This is the most anticipated earnings report since NVIDIA's Q1 FY26 beat that kicked off the second leg of the AI rally. The AMD-Meta deal provides a demand-side validation but also raises competitive margin questions. If NVIDIA confirms mid-70s gross margins, the S&P likely pushes through the 200D. If margins disappoint, the convergence zone between the 50D and 200D resolves lower. Results expected ~4:20pm ET, conference call at 5pm.
Current state: Third round of Geneva talks Thursday. Iran ruling out interim deal. US military buildup largest since 2003. Today's update: Trump's SOTU offered nothing new — "I would never let Iran have a nuclear weapon" but no clarity on the military option. Araghchi's claim about zero enrichment not being demanded was not addressed. LSE professor Fawaz Gerges suggested Trump may strike Iran to "divert attention" from the SCOTUS tariff defeat. Oil at $66 is pricing diplomacy; if talks collapse Thursday, the risk premium reprices immediately. Gold above $5,180 is already hedging the downside scenario.
Current state: IEEPA tariffs struck down 6-3 by SCOTUS (Feb 20). Section 122 replacement: 10% implemented, 15% announced but not formalized. Today's update: The legal vulnerability is worsening. Trump's own DOJ previously said Section 122 doesn't apply to the current situation. Former IMF Deputy Gita Gopinath publicly rejected the "balance of payments crisis" justification. Litigation is inevitable and the timeline is short. $160B in collected IEEPA tariffs with no refund framework. EU froze trade deal ratification. Japan seeking assurances on its $550B investment pledge. The tariff regime is simultaneously operative and legally incoherent — businesses can't model costs.
Current state: Software ETFs down 17-23% YTD. Monday's Anthropic catalyst gave the thesis its first live demonstration. Today's update: The Spotify data point extends the narrative. If a $100B company's best engineers are now supervisors of AI output rather than creators, the labor cost assumptions embedded in every SaaS valuation model need revision. This isn't a one-day event — it's the beginning of a structural repricing that runs through the entire earnings season. Watch for "AI efficiency" commentary in every tech earnings call this quarter.
Current state: Gold $5,183. Silver $88. Goldman $5,400 year-end target. Gold/DXY decoupling sustained. Today's update: Gold continues to climb on every risk channel simultaneously: tariff chaos, Iran, dollar weakness, equity volatility. Fourth straight week of gains. China's central bank extended gold purchases for a fifteenth consecutive month. Silver at $88, up $55 from a year ago — nominal ATH was $121.67 set January 29. 10Y yield holding near 4%, closest to sub-4% since late November. CME data: 98% expect rates unchanged at 3.50-3.75%. When rates are flat, the dollar is falling, and geopolitical risk is rising from multiple directions, gold has no natural ceiling.
Current state: BTC ~$65,500. ETH ~$1,915. SOL ~$82. All deeply below 50D and 200D moving averages. BTC 48% below ATH. Today's update: BTC bounced from $62,858 to ~$65,500 — a 4% relief rally on light volume. US demand negative for a record 40 consecutive days. Vitalik sold 17K ETH this month. Total market cap $2.33T. The acute selling has paused but the bear structure is intact: lower highs, institutional outflows, no catalyst for reversal. If BTC can't hold $65K, the $60K test is next, and below that the 200W MA at ~$42K becomes the conversation.
Current state: El Mencho killed with US intelligence support. CJNG retaliated — 25 National Guard killed, roadblocks, arson. Airlines canceling Mexico flights. Shelter in place for US citizens. Today's update: No change from yesterday's update. The violence is ongoing. Supply chain impact to US-Mexico trade being monitored. Airline cancellations continue for Puerto Vallarta and Guadalajara routes.
Current state: FEMA at $9.6B (from $30B). Global Entry suspended. TSA workers miss first full paycheck March 14 if unresolved. Today's update: SOTU didn't produce a deal. Congress is back from recess but no resolution timeline. The shutdown is now the longest DHS-specific funding gap in history. If it extends past March 14, the public impact escalates significantly.
Remaining Big Stories — no change today: The Fed's Impossible Position, AI Capex Cycle (updated under NVIDIA above), Humanoid Robotics, Crypto Regulatory Clarity, India Energy Realignment, Political Violence Trend, US-China Tech Decoupling, Nuclear Renaissance, Strategy BTC Treasury Risk, Silver Supply Deficit, AI Model Architecture Shift, Japan Monetary Policy, European Defense Spending, US Fiscal Trajectory, Global Dollar System.
# ▸ TOMORROW'S HEADLINES What the market will be talking about in 12-18 months.
New evidence today: Spotify's most senior developers haven't written a line of code since December. They review and supervise AI output. If this is reproducible across the industry — and the economics suggest it is — the ratio of engineers to output is about to change dramatically. The first-order effect: fewer engineering hires. The second-order effect: engineering management becomes the bottleneck, not engineering execution. The third-order effect: companies built on "we have the best engineers" lose their moat when the tool matters more than the operator.
New evidence today: If NVIDIA reports mid-70s gross margins tonight, it validates that AI infrastructure is both high-growth and high-margin — the rarest combination in tech. If margins compress, even on a revenue beat, it signals that competition (AMD-Meta) and mix shift (Blackwell ramp) are beginning to erode pricing power. The margin trajectory tonight sets the narrative for every AI infrastructure company through the rest of 2026.
Full reference list (20 items) unchanged — see below.
# ▸ THE WATCHLIST Regime-changing 2-10x opportunities. Small bets, big asymmetry. Most will be wrong.
This section is purely illustrative — not investment advice. These are structural theses applied to specific assets to test our frameworks against real markets. Do not invest in anything because it appears here. Do your own work. Size accordingly.
ETN — Software company disguised as an industrial ~$380. Data center orders +200%. Electrical backlog $15.3B. The insight: Every data center needs recurring power management software that isn't broken out. Re-rates from "industrial 29x" to "hybrid 35x+." If the AI-energy convergence plays out, 3-5x over 3-5 years. Upside: Software reclassification at 35x = $560 (47%). AI energy narrative = $700+ (80%+). Downside: Buildout slows, range-bound (~10%). Validates: Software-attached revenue reported. Data center >25% of revenue. Rejects: Margins flat. No software breakout through 2027. Feb 25 note: NVIDIA earnings tonight are the next data point. If Blackwell ramp confirms the capacity buildout is accelerating, every data center power management thesis gets pulled forward. ETN's backlog is front-of-line for the 6GW of AMD GPUs Meta just committed to.
SOL — Agent payment rails ~$82, down 76% from ATH. The insight: AI agents need micropayments at machine speed. Banks can't clear $0.001. Solana can. Leading all chains in payment volume during a bear. Staking ETFs creating institutional floor. Nobody models machine transaction volume. Upside: Agent commerce + cycle recovery = $200-300 (2.5-3.5x). Machine commerce real = $500+ (6x). Downside: Agents use Stripe. SOL to $40 (~50%). Validates: Agent tx volume on Solana. x402 >$1B cumulative. ETF inflow trend persists. Rejects: Agent commerce centralizes. Reliability issues. Feb 25 note: SOL range-bound at $82 for two weeks in the $75-$85 zone. Deeply negative funding rates + persistent ETF inflows = the contrarian setup remains intact. If macro stabilizes, this has the most explosive upside of the three majors. If not, $50 is the next test. The agent commerce thesis doesn't depend on the strategic reserve narrative — but the narrative is providing a sentiment floor.
Harmonic Drive (6324.T) — Humanoid picks-and-shovels ~¥3,800. The insight: Every humanoid platform needs the same precision actuators. Near-monopoly on strain wave gears, minimal Western coverage. Wins regardless of which platform wins. Upside: 500K units/yr by 2028, 20-40 actuators each. Revenue triples = 3x+. Downside: Timeline slips. Chinese competitors. 30-50% down. Validates: Tesla ships externally. Humanoid OEM backlog growth. Rejects: Production <10K in 2026.
Gulf Sovereign AI — Thematic, watching for vehicle The insight: Gulf running oil playbook with compute. Cheapest energy, 3.5B people need AI outside US/China. 95% of AI allocation is US tech. UAE at 14x vs S&P 21x. Upside: G42 IPO creates investable category. 30%+ broad. Downside: Captured by hyperscalers. 10-15%. Validates: G42 IPO. Sovereign AI deals bypass AWS/Azure. Rejects: Hyperscalers go direct. No differentiated tech.
The Thermodynamic Cost of Learning Is Real — and the AI Industry Is About to Hit It — Gartner projects $2.5 trillion in AI spending for 2026, a 44% jump from 2025. Here's the physics underneath that number: every computation generates heat, every learning cycle increases entropy, every GPU cluster requires cooling. Information theory tells us there's no free lunch — gaining information requires energy and produces waste heat as a thermodynamic byproduct. This is why the AMD-Meta deal is measured in gigawatts, not teraflops. The constraint on AI scaling isn't algorithmic anymore — it's thermodynamic. The companies solving the energy-to-computation conversion most efficiently (nuclear, advanced cooling, power management) may ultimately capture more value than the model builders. When the limiting factor shifts from software to physics, the investment thesis shifts with it.
Current state: Report due after close today. $65.7B revenue consensus, $1.53 EPS. Blackwell ramp, China sales, and Q1 guidance are the three variables that matter.
Today's update: The setup is maximally crowded — 94.5% beat probability on Polymarket, 37 buy / 1 hold / 1 sell from analysts. This is the most anticipated earnings report since NVIDIA's Q1 FY26 beat that kicked off the second leg of the AI rally. The AMD-Meta deal provides a demand-side validation but also raises competitive margin questions. If NVIDIA confirms mid-70s gross margins, the S&P likely pushes through the 200D. If margins disappoint, the convergence zone between the 50D and 200D resolves lower. Results expected ~4:20pm ET, conference call at 5pm.
Current state: Third round of Geneva talks Thursday. Iran ruling out interim deal. US military buildup largest since 2003.
Today's update: Trump's SOTU offered nothing new — "I would never let Iran have a nuclear weapon" but no clarity on the military option. Araghchi's claim about zero enrichment not being demanded was not addressed. LSE professor Fawaz Gerges suggested Trump may strike Iran to "divert attention" from the SCOTUS tariff defeat. Oil at $66 is pricing diplomacy; if talks collapse Thursday, the risk premium reprices immediately. Gold above $5,180 is already hedging the downside scenario.
Current state: IEEPA tariffs struck down 6-3 by SCOTUS (Feb 20). Section 122 replacement: 10% implemented, 15% announced but not formalized.
Today's update: The legal vulnerability is worsening. Trump's own DOJ previously said Section 122 doesn't apply to the current situation. Former IMF Deputy Gita Gopinath publicly rejected the "balance of payments crisis" justification. Litigation is inevitable and the timeline is short. $160B in collected IEEPA tariffs with no refund framework. EU froze trade deal ratification. Japan seeking assurances on its $550B investment pledge. The tariff regime is simultaneously operative and legally incoherent — businesses can't model costs.
Current state: Software ETFs down 17-23% YTD. Monday's Anthropic catalyst gave the thesis its first live demonstration.
Today's update: The Spotify data point extends the narrative. If a $100B company's best engineers are now supervisors of AI output rather than creators, the labor cost assumptions embedded in every SaaS valuation model need revision. This isn't a one-day event — it's the beginning of a structural repricing that runs through the entire earnings season. Watch for "AI efficiency" commentary in every tech earnings call this quarter.
Current state: Gold $5,183. Silver $88. Goldman $5,400 year-end target. Gold/DXY decoupling sustained.
Today's update: Gold continues to climb on every risk channel simultaneously: tariff chaos, Iran, dollar weakness, equity volatility. Fourth straight week of gains. China's central bank extended gold purchases for a fifteenth consecutive month. Silver at $88, up $55 from a year ago — nominal ATH was $121.67 set January 29. 10Y yield holding near 4%, closest to sub-4% since late November. CME data: 98% expect rates unchanged at 3.50-3.75%. When rates are flat, the dollar is falling, and geopolitical risk is rising from multiple directions, gold has no natural ceiling.
Current state: BTC ~$65,500. ETH ~$1,915. SOL ~$82. All deeply below 50D and 200D moving averages. BTC 48% below ATH.
Today's update: BTC bounced from $62,858 to ~$65,500 — a 4% relief rally on light volume. US demand negative for a record 40 consecutive days. Vitalik sold 17K ETH this month. Total market cap $2.33T. The acute selling has paused but the bear structure is intact: lower highs, institutional outflows, no catalyst for reversal. If BTC can't hold $65K, the $60K test is next, and below that the 200W MA at ~$42K becomes the conversation.
Current state: El Mencho killed with US intelligence support. CJNG retaliated — 25 National Guard killed, roadblocks, arson. Airlines canceling Mexico flights. Shelter in place for US citizens.
Today's update: No change from yesterday's update. The violence is ongoing. Supply chain impact to US-Mexico trade being monitored. Airline cancellations continue for Puerto Vallarta and Guadalajara routes.
Current state: FEMA at $9.6B (from $30B). Global Entry suspended. TSA workers miss first full paycheck March 14 if unresolved.
Today's update: SOTU didn't produce a deal. Congress is back from recess but no resolution timeline. The shutdown is now the longest DHS-specific funding gap in history. If it extends past March 14, the public impact escalates significantly.
Remaining Big Stories — no change today: The Fed's Impossible Position, AI Capex Cycle (updated under NVIDIA above), Humanoid Robotics, Crypto Regulatory Clarity, India Energy Realignment, Political Violence Trend, US-China Tech Decoupling, Nuclear Renaissance, Strategy BTC Treasury Risk, Silver Supply Deficit, AI Model Architecture Shift, Japan Monetary Policy, European Defense Spending, US Fiscal Trajectory, Global Dollar System.
New evidence today: Spotify's most senior developers haven't written a line of code since December. They review and supervise AI output. If this is reproducible across the industry — and the economics suggest it is — the ratio of engineers to output is about to change dramatically. The first-order effect: fewer engineering hires. The second-order effect: engineering management becomes the bottleneck, not engineering execution. The third-order effect: companies built on "we have the best engineers" lose their moat when the tool matters more than the operator.
New evidence today: If NVIDIA reports mid-70s gross margins tonight, it validates that AI infrastructure is both high-growth and high-margin — the rarest combination in tech. If margins compress, even on a revenue beat, it signals that competition (AMD-Meta) and mix shift (Blackwell ramp) are beginning to erode pricing power. The margin trajectory tonight sets the narrative for every AI infrastructure company through the rest of 2026.
This section is purely illustrative — not investment advice. These are structural theses applied to specific assets to test our frameworks against real markets. Do not invest in anything because it appears here. Do your own work. Size accordingly.
ETN — Software company disguised as an industrial
~$380. Data center orders +200%. Electrical backlog $15.3B.
The insight: Every data center needs recurring power management software that isn't broken out. Re-rates from "industrial 29x" to "hybrid 35x+." If the AI-energy convergence plays out, 3-5x over 3-5 years.
Upside: Software reclassification at 35x = $560 (47%). AI energy narrative = $700+ (80%+).
Downside: Buildout slows, range-bound (~10%).
Validates: Software-attached revenue reported. Data center >25% of revenue.
Rejects: Margins flat. No software breakout through 2027.
Feb 25 note: NVIDIA earnings tonight are the next data point. If Blackwell ramp confirms the capacity buildout is accelerating, every data center power management thesis gets pulled forward. ETN's backlog is front-of-line for the 6GW of AMD GPUs Meta just committed to.
SOL — Agent payment rails
~$82, down 76% from ATH.
The insight: AI agents need micropayments at machine speed. Banks can't clear $0.001. Solana can. Leading all chains in payment volume during a bear. Staking ETFs creating institutional floor. Nobody models machine transaction volume.
Upside: Agent commerce + cycle recovery = $200-300 (2.5-3.5x). Machine commerce real = $500+ (6x).
Downside: Agents use Stripe. SOL to $40 (~50%).
Validates: Agent tx volume on Solana. x402 >$1B cumulative. ETF inflow trend persists.
Rejects: Agent commerce centralizes. Reliability issues.
Feb 25 note: SOL range-bound at $82 for two weeks in the $75-$85 zone. Deeply negative funding rates + persistent ETF inflows = the contrarian setup remains intact. If macro stabilizes, this has the most explosive upside of the three majors. If not, $50 is the next test. The agent commerce thesis doesn't depend on the strategic reserve narrative — but the narrative is providing a sentiment floor.
Harmonic Drive (6324.T) — Humanoid picks-and-shovels
~¥3,800.
The insight: Every humanoid platform needs the same precision actuators. Near-monopoly on strain wave gears, minimal Western coverage. Wins regardless of which platform wins.
Upside: 500K units/yr by 2028, 20-40 actuators each. Revenue triples = 3x+.
Downside: Timeline slips. Chinese competitors. 30-50% down.
Validates: Tesla ships externally. Humanoid OEM backlog growth.
Rejects: Production <10K in 2026.
Gulf Sovereign AI — Thematic, watching for vehicle
The insight: Gulf running oil playbook with compute. Cheapest energy, 3.5B people need AI outside US/China. 95% of AI allocation is US tech. UAE at 14x vs S&P 21x.
Upside: G42 IPO creates investable category. 30%+ broad.
Downside: Captured by hyperscalers. 10-15%.
Validates: G42 IPO. Sovereign AI deals bypass AWS/Azure.
Rejects: Hyperscalers go direct. No differentiated tech.
# ▸ DISCOVERY
The Thermodynamic Cost of Learning Is Real — and the AI Industry Is About to Hit It — Gartner projects $2.5 trillion in AI spending for 2026, a 44% jump from 2025. Here's the physics underneath that number: every computation generates heat, every learning cycle increases entropy, every GPU cluster requires cooling. Information theory tells us there's no free lunch — gaining information requires energy and produces waste heat as a thermodynamic byproduct. This is why the AMD-Meta deal is measured in gigawatts, not teraflops. The constraint on AI scaling isn't algorithmic anymore — it's thermodynamic. The companies solving the energy-to-computation conversion most efficiently (nuclear, advanced cooling, power management) may ultimately capture more value than the model builders. When the limiting factor shifts from software to physics, the investment thesis shifts with it.
Proposed changes based on today's brief:
NVIDIA Earnings elevated to Big Story #1 for the day. The most anticipated earnings report in the market. Results after close tonight will determine whether the AI capex thesis re-accelerates or faces its first real test. Moves to normal tracking after the event.
Tariff Regime — legal vulnerability increasing. Section 122 is now under active challenge with Trump's own DOJ having previously said it doesn't apply. The $160B refund question is unresolved. The effective tariff regime is lower than pre-SCOTUS but legally less stable.
State of the Union — no market-moving signals. Trump offered nothing new on Iran, tariffs, or the shutdown. The speech was defensive, not programmatic. 60% disapproval going in. The market correctly priced it as a non-event.
Gold safe haven bid strengthening. $5,183 with silver surging to $91. China PBOC buying for 15th consecutive month. When every macro risk channel feeds gold simultaneously and there's no offsetting catalyst, the trend accelerates.
Tomorrow's Headlines addition: Engineer-to-Supervisor transition. Spotify data point is the most concrete evidence yet that AI is changing the nature of engineering work at scale. Not hypothetical — deployed and running at a $100B company.
VIX easing but still elevated. Down from Monday's 21+ spike to 19.55 on Tuesday — a 7% decline but still well above the ~14 average. The fear has moderated, not resolved. NVIDIA earnings tonight are the next volatility catalyst. A clean beat likely pushes VIX back toward 16-17. A margin or guidance miss could send it back above 21.
Source check: NVIDIA earnings call transcript (tonight) should be added to tomorrow's Deep Read. Tax Foundation's SCOTUS analysis is the best legal resource on the tariff situation.
Third round of Geneva talks Thursday. Iran ruling out interim deal. Trump's SOTU offered no clarity. Military buildup remains largest since 2003. Oil at $66. LSE: Trump may strike to divert from SCOTUS loss.
Updated Feb 25.
Anthropic catalysts digesting. Spotify's senior engineers stopped writing code in December — supervise AI output. The labor cost assumptions in SaaS valuations need revision. Software ETFs down 17-23% YTD.
Updated Feb 25.
BTC ~$65,500, ETH ~$1,915, SOL ~$82. BTC bounced from $62,858. US demand negative 40 consecutive days. Vitalik sold 17K ETH in Feb. Total market cap $2.33T. Bear structure intact — all assets below 50D/200D MAs.
Updated Feb 25.
Gold $5,183. Silver $88. Fourth straight week of gains. China PBOC buying 15th consecutive month. Goldman $5,400 year-end target. 10Y near 4% — closest to sub-4% since Nov. Every macro risk channel feeding the bid.
Updated Feb 25.
Rates 3.50-3.75%. Collins: holding steady "likely appropriate." Barkin: well-positioned. Markets pricing three cuts. 98% expect unchanged in March. Warsh takes chair May. Next test: Feb labor (March 6) + CPI (March 11).
Updated Feb 25.
IEEPA tariffs struck down 6-3 (Feb 20). Section 122: 10% implemented, 15% announced. DOJ previously said Section 122 doesn't apply. $160B collected with no refund framework. Former IMF Deputy rejected balance of payments justification. EU froze trade deal. Litigation imminent.
Updated Feb 25.
$650B+ hyperscaler capex in 2026. AMD-Meta $100B deal validated spending acceleration. NVIDIA Q4 earnings after close: $65.7B consensus. Blackwell margin trajectory and Q1 guide are the tests. Gartner: $2.5T total AI spending in 2026 (+44%).
Updated Feb 25.
Tesla Gen 3 mass production Jan 2026. Figure at BMW. 1X at $20K consumer. Target 100K+ units 2026. Watching GTC March 16.
Last updated Feb 20.
GENIUS Act regs due July 18. CLARITY Act pushing for Senate vote. FDIC approved bank stablecoin issuance. Grayscale AAVE ETF filed.
Last updated Feb 20.
India's potential pivot away from Russian oil (1.8M bpd) linked to US trade deal. Trade talks paused post-SCOTUS.
Last updated Feb 23.
Day 11. FEMA at $9.6B (from $30B). Global Entry suspended. SOTU produced no deal. Congress back from recess. TSA workers miss paycheck March 14 if unresolved. Longest DHS-specific shutdown in history.
Updated Feb 25.
Mar-a-Lago breach, Capitol breach, accelerating pattern. Capitol Police: 14,938 cases in 2025 vs 9,474 in 2024.
Last updated Feb 23.
Export controls tightening. Two separate AI ecosystems forming. NVIDIA H200 China sales permitted with 25% revenue share to US government.
Updated Feb 25.
AI power demand driving nuclear restart. AMD-Meta 6GW deal underscores the scale. AI spending hitting thermodynamic constraints.
Updated Feb 25.
717K BTC at $76K avg. BTC at ~$65.5K = ~$7.5B unrealized loss. Convertible structure risk elevated. Every dollar lower increases forced selling risk.
Updated Feb 25.
6th consecutive year of structural deficit (67M oz). Silver at $88, up $55 YoY. Nominal ATH $121.67 (Jan 29). Industrial + monetary demand converging.
Updated Feb 25.
Training paradigm showing diminishing returns. Inference-time compute emerging as new frontier. Spotify: senior engineers supervising AI output, not writing code.
Updated Feb 25.
BOJ exiting zero rates. Largest foreign holder of US Treasuries. Yen weakening despite intervention speculation. Japan seeking tariff assurances on $550B deal.
Updated Feb 25.
NATO budgets rising. Fiscal pressure on stretched European sovereigns.
Last updated Feb 22.
$36T+ debt. Interest exceeding defense spending. Fiscal dominance dynamics intensifying.
Last updated Feb 22.
DXY at 97.66 — down 8% YoY. Central bank gold buying as dollar diversification proxy. Three consecutive months of decline before minor February bounce.
Updated Feb 25.
El Mencho (CJNG leader) killed with US intelligence support. Immediate cartel retaliation: 25 National Guard killed, roadblocks, arson. Airlines canceling Mexico flights. US citizens shelter in place. Violence ongoing.
Updated Feb 25.
"Who has the power" > "who has the chips." Nuclear renaissance, grid infrastructure, energy-sovereign AI. *Evidence Feb 25: AMD-Meta deal measured in gigawatts. $2.5T AI spending hitting thermodynamic constraints.*
x402, Coinbase Wallets, Lightspark. Machine-speed settlement. *Evidence Feb 23: BTC Sunday crash shows why 24/7 settlement matters.*
$1.2T globally. Pure labor arbitrage. *Evidence Feb 24: IBM -13% on AI capability demo is the Phase 3 template for consulting.*
IBM + Microsoft converging. First commercial quantum application within 18 months.
Gulf positioning first. India and Brazil next. *Evidence Feb 25: India hosted 5-day AI summit in New Delhi on AI governance and sovereignty.*
HBM4 supply-constrained. SK Hynix/Samsung bottleneck. AI SSDs emerging.
Inference energy economics could be transformative.
GENIUS Act Jan 2027. Banks can issue. *Evidence Feb 23: ProShares ETF debuted $17B.*
Built-on-AI > bolt-on-AI. *Evidence Feb 24: AMD-Meta (native) vs IBM (augmented) divergence in a single session.*
If open-source reaches parity, value shifts to application layer.
Privacy, latency, cost advantages.
Pay-per-task robot labor. $15/hr equivalent vs $25/hr human.
Biology as manufacturing platform. AI + protein design accelerating.
Voluntary + compliance converging. Blockchain verification.
Proof of humanity becomes real need. Deepfakes + agents driving urgency.
GLP-1 mortality data. CRISPR approved. Anti-aging Phase 3.
Desalination improving. Moving from ESG niche to core infra.
Starship 10x cost reduction. $469B → $1T+ by 2030.
On-chain risk transfer. $6T traditional insurance market.
AI displacing white-collar faster than blue-collar. Political economy of displacement.
New evidence Feb 25: Spotify engineers now supervise AI code, not write it. The transition is happening.