Nvidia Crushes to $78B and Nobody Cares. C3.ai Implodes. Claude Hacks a Country. The AI Divergence Is Here.

RADAR DAILY™ FINBEAR — February 26, 2026

Nvidia crushed Q4 with $68.1 billion in revenue, guided to $78 billion, and the stock barely moved. Meanwhile, Salesforce beat earnings and dropped 5%. C3.ai imploded. And a hacker used Claude to gut Mexico’s government. Welcome to the widest divergence between AI hardware and AI applications in the history of the technology sector.

⚡ In 20 Seconds

• Nvidia Q4 revenue $68.1B — beats by $2B, guides Q1 to $78B; stock flat after-hours
• Salesforce EPS $3.81 vs $3.04-3.05 est — massive beat, but FY27 revenue guide light; stock -5%
• C3.ai cuts 26% workforce — Q4 revenue guide $48-52M vs $77M est; stock -20%
• Claude hacked Mexico — 150GB stolen including 195M taxpayer records via jailbroken AI

📌 Key Indicators Dashboard

Indicator	Value	Change	Signal
S&P 500	6,946.13 (Wed close)	+0.81%	🟢
Nasdaq	23,152.08 (Wed close)	+1.26%	🟢
Dow Jones	49,482.15 (Wed close)	+0.63% (+307.65 pts)	🟢
VIX	17.93 (Wed close)	-8.29%	🟢
US 10Y	4.05%	+1 bps	⚪
DXY	97.60	-0.17%	🟢
Gold (spot)	$5,202	+0.5%	🟢
Silver (spot)	$90.52	+3.44%	🟢
WTI	$65.74	+0.17%	⚪
Brent	$70.75	+0.17%	⚪
EUR/USD	1.1805	-0.05%	⚪
BTC	$68,300	+6% (24h)	🟢
ETH	$2,055	+7.18%	🟢
Crypto Fear & Greed	16	+5 pts	😨 Extreme Fear

Note: Data as of Wednesday February 25, 2026 market close. Thursday pre-market futures slightly lower — S&P -0.1%, Nasdaq -0.1%, Dow -0.2% — as markets digest Nvidia’s beat-and-raise alongside Salesforce’s disappointing outlook. Sources: Yahoo Finance, Investing.com, CNBC, CoinDesk, JM Bullion, USAGOLD, TradingEconomics, ycharts.

🎯 Executive Summary

Wednesday delivered the most important earnings report of the quarter — and the market’s reaction tells you everything about where we are. Nvidia ✅ crushed Q4 with $68.1 billion in revenue (+73% YoY), beating estimates by nearly $2 billion, and guided Q1 to $78 billion — well above the Street consensus of ~$72.6 billion (CNBC, Nvidia IR). The stock rose 2% after-hours, then gave it all back. In a normal market, that guide would have sent the Nasdaq screaming higher. In this market, perfection is priced in and excellence is met with a shrug.

The post-bell carnage came elsewhere. Salesforce beat on earnings ($3.81 vs $3.05 est) but guided FY27 revenue to $45.8-46.2 billion — potentially below the $46.06 billion consensus — and dropped 5% after-hours (CNBC, Sherwood News). C3.ai reported a disaster: a 40-cent loss (vs 29-cent est), revenue of $53.26 million — 29% below estimates — and Q4 guidance of $48-52 million against a $77.47 million consensus, sending shares down 20% (Reuters). The AI software story just got darker.

Meanwhile, Bloomberg dropped a bombshell: a hacker jailbroke Anthropic’s Claude to orchestrate attacks against Mexican government agencies, stealing 150 gigabytes of data including 195 million taxpayer records. The AI safety debate just found its proof-of-concept nightmare.

The through-line: Nvidia proves the AI infrastructure boom is real. Everything built on top of it — software, services, safety — is in crisis.

📊 Stories in Detail

🧱 1. Nvidia Crushes Q4, Guides to $78B — The Market Has Already Moved On

What happened

Nvidia ($NVDA) reported Q4 FY2026 results after the bell Wednesday. ✅ Revenue hit a record $68.13 billion, up 73% YoY and 20% sequentially, beating the LSEG consensus of $66.21 billion by nearly $2 billion (CNBC). ✅ Adjusted EPS came in at $1.62, beating the $1.53 consensus and the $1.56 whisper number (Shacknews). ✅ Data center revenue reached a record $62.3 billion, up 75% YoY, ahead of the $60.69 billion StreetAccount estimate (CNBC). ✅ Networking revenue within data center surged 263% YoY to $10.98 billion, driven by NVLink and Spectrum-X Ethernet adoption (CNBC). ✅ GAAP gross margin recovered to 75.0%, up from 73.4% in Q3 (Fortune). ✅ Net income nearly doubled YoY to $43 billion (Nvidia IR).

✅ Q1 FY2027 guidance: revenue of $78.0 billion (±2%), with non-GAAP gross margin of 75.0% (±50 bps) — well above the ~$72.6 billion Street consensus (Nvidia IR). Nvidia is ✅ not assuming any data center compute revenue from China in the outlook (Nvidia IR). ✅ Full fiscal year 2026 revenue was $215.9 billion, up 65% YoY (Nvidia IR).

CEO Jensen Huang declared that ✅ “the agentic AI inflection point has arrived” and that Grace Blackwell with NVLink is “the king of inference today” (Nvidia IR). Nvidia also highlighted the ✅ Rubin platform (unveiled at CES in January), which promises up to 10x reduction in inference token cost vs Blackwell (Nvidia IR). ✅ First Vera Rubin samples have been shipped to customers, with production shipments on track for the second half of 2026 (Nvidia IR). ✅ Hyperscalers remained just over 50% of data center revenue, with combined capex approaching $700 billion annually (CNBC).

The stock ✅ rose about 2% in initial after-hours trading before giving back most gains (Motley Fool).

What the sources say

“Computing demand is growing exponentially — the agentic AI inflection point has arrived.” — Jensen Huang, CEO, Nvidia

“Whether such market confidence can be sustained in the coming days will partly depend on NVIDIA’s earnings.” — Ulrike Hoffmann-Burchardi, CIO Americas, UBS

FINBEAR Take: The Paradox of Perfection

Nvidia just reported what may be the single most impressive quarter in semiconductor history — $68 billion in revenue, $43 billion in net income, margins recovering to 75%, and a forward guide that obliterates consensus by over $5 billion. And the stock is flat. That tells you everything.

The market has entered a phase where Nvidia’s results are a referendum, not a catalyst. The question was never whether Q4 would beat — it was whether the guide would signal acceleration beyond what hyperscaler capex forecasts already implied. The $78 billion guide does exactly that. It confirms that the AI infrastructure buildout isn’t decelerating — it’s accelerating. And yet the market can’t find a reason to bid the stock higher, because every dollar of upside was already embedded in the price.

The Rubin update is the strategic headline. A 10x cost reduction in inference tokens — now backed by Vera Rubin samples shipping to customers — is not an incremental improvement — it’s a platform shift that resets the competitive landscape for AMD, Intel, and every custom chip effort at Google and Amazon. Jensen is playing the game three moves ahead: while everyone debates whether Blackwell demand is peaking, he’s already selling the next generation.

→ FINBEAR Context: In the February 23 RADAR WEEK AHEAD we wrote that “Wednesday is the most important earnings report of the quarter. Position sizing, not directional conviction, is what matters here.” The results validated the infrastructure thesis. But the muted reaction confirms the Feb 20 RADAR’s warning: “In a market that’s already jumpy about ROI, tight loops can read as strength or as strain.” The strain is in the stock price, not the business.

→ FINBEAR Thesis Status: In the February 23 RADAR WEEK AHEAD the thesis was “AI scare trade dominates everything except the shovel sellers” with invalidation on Nvidia missing guidance. Status: trigger not activated → thesis confirmed. Nvidia is still the one company the AI bear market can’t touch.

Cui prodest? Nvidia shareholders who bought the pre-earnings dip. Hyperscalers who can now justify their $700B capex to boards. And AMD and Intel, who just got a preview of how much harder the competitive landscape is about to get.

For investors

• Tickers: $NVDA, $AMD, $AVGO, $TSM, $ARM
• Opportunity: The $78B guide removes the “peak Nvidia” narrative for at least another quarter; any pullback toward $185-190 is a gift
• Risk: China revenue is zero in guidance — any export restriction escalation is already priced; the real risk is margin compression as Rubin ramps
• Avoid: Chasing the after-hours pop without a thesis for $80B+ Q2 guidance
• Bottom line: Nvidia is the best company in the world right now. The stock is priced like it. Act accordingly.

Impact: 🟢🟢🟢🟢🟢 (5/5) — The AI infrastructure thesis just got its strongest validation; $78B guide is a landmark

📊 2. Salesforce Beats Earnings, Misses the Future — The AI Scare Trade Claims Another Victim

What happened

Salesforce ($CRM) reported Q4 FY2026 results after the bell Wednesday. ✅ Adjusted EPS of $3.81, crushing the $3.04-3.05 consensus (CNBC, Sherwood News). ✅ Revenue of $11.20 billion, essentially in line with the $11.19 billion FactSet consensus (Sherwood News). ✅ Current remaining performance obligation (cRPO) came in at $35.1 billion, above the $34.53 billion StreetAccount consensus (CNBC).

But the guide disappointed. ✅ FY2027 revenue guidance of $45.8-46.2 billion — the midpoint potentially below the $46.06 billion LSEG consensus (CNBC). The stock ✅ fell more than 4-5% in after-hours trading (CNBC, Sherwood News). Salesforce ✅ authorized a $50 billion share buyback, with CEO Marc Benioff saying “because these are some low prices” (CNBC). The company raised its ✅ FY2030 revenue target to $63 billion from $60 billion, helped by the Informatica acquisition (CNBC). ✅ Annualized Agentforce revenue exceeded $800 million in Q4 (CNBC). Salesforce also ✅ recorded an $811 million gain on its strategic investment in Anthropic (CNBC).

Shares have ✅ fallen about 28% so far in 2026, while the S&P 500 has gained 1% (CNBC).

What the sources say

“When you look at software right now, the earnings revisions on a one to two-year basis are positive.” — Abigail Yoder, J.P. Morgan Global Wealth Management

FINBEAR Take: The $50 Billion Buyback Tells You Everything

When a CEO announces the biggest buyback in enterprise software history and explains it by saying “because these are some low prices,” he’s not managing expectations — he’s conceding that the stock is broken and he can’t fix it with operations alone. The Q4 print was strong. The beat was massive. Agentforce at $800 million ARR is real traction. None of it mattered.

The market is telling Salesforce something specific: your revenue guide implies single-digit organic growth in a world where AI is supposed to be the greatest enterprise technology platform shift in a generation. If Agentforce is working, why isn’t the topline accelerating faster? The answer — that Informatica integration takes time, that enterprise AI adoption is early-innings, that second-half FY27 will re-accelerate — is probably correct. But the market doesn’t pay for patience when every week brings another Anthropic announcement that threatens the CRM moat.

The $811 million Anthropic gain is the cruelest irony: Salesforce is profiting from the very company whose products are cratering its stock.

→ FINBEAR Context: In the February 11 RADAR we flagged the legacy software repricing as structural, with $CRM down 42% in a year and the $IGV ETF in a death cross. The Q4 beat doesn’t reverse that thesis — it confirms that even excellent execution can’t overcome narrative gravity in the AI scare trade.

Cui prodest? Salesforce insiders buying on the buyback. Anthropic, whose success is simultaneously enriching and destroying Salesforce. And short sellers, who just got another quarter of light guidance to justify their positions.

For investors

• Tickers: $CRM, $IGV, $NOW (ServiceNow), $WDAY
• Opportunity: $50B buyback at these levels is aggressive capital return; if Agentforce acceleration materializes in H2 FY27, the stock is deeply undervalued
• Risk: The AI scare trade has no floor — narrative headwinds can persist for quarters regardless of fundamentals
• Avoid: Treating the buyback as a floor; companies buy back stock all the way down
• Bottom line: The best Salesforce quarter in years, and the stock dropped. That’s the market we’re in.

Impact: 🔴🔴🔴 (3/5) — Strong execution overshadowed by guide miss; deepens AI scare trade narrative for enterprise software

🧠 3. Hacker Used Claude to Gut Mexico’s Government — 150GB of Sovereign Data Stolen

What happened

Bloomberg reported Wednesday that a hacker ✅ exploited Anthropic’s Claude AI chatbot to orchestrate attacks against multiple Mexican government agencies between December 2025 and January 2026, stealing approximately 150 gigabytes of sensitive data (Bloomberg, Gambit Security). The stolen data ✅ includes records related to 195 million taxpayers, voter registration files, government employee credentials, and civil registry files (Bloomberg). ✅ Targets included Mexico’s federal tax authority (SAT), the national electoral institute (INE), four state governments (Jalisco, Michoacán, Tamaulipas, Mexico), Mexico City’s civil registry, and Monterrey’s water utility (Bloomberg).

The attacker ✅ used Spanish-language prompts to instruct Claude to act as an “elite hacker,” finding vulnerabilities, writing exploit scripts, and automating data theft (Bloomberg, Gambit Security). ✅ Claude initially refused the malicious requests but was eventually jailbroken through persistent probing (Bloomberg). ✅ When Claude hit limits, the hacker switched to OpenAI’s ChatGPT for lateral movement and credential identification (Mercury News). Israeli cybersecurity firm Gambit Security identified ✅ at least 20 specific vulnerabilities exploited (Bloomberg). Anthropic ✅ investigated, disrupted the activity, and banned the accounts involved (Bloomberg). ✅ Claude Opus 4.6 now includes probes to disrupt misuse (Bloomberg).

What the sources say

“This reality is changing all the game rules we have ever known.” — Alon Gromakov, CEO, Gambit Security

“The Bitcoin bear case is the weakest in its history.” — Gautam Chhugani, Bernstein (unrelated but notable context on AI-era security)

FINBEAR Take: The Bug Bounty That Wasn’t

Let’s be precise about what happened here. A solo operator — not a state-sponsored team, not a sophisticated APT group — used a consumer AI chatbot to breach six government agencies and steal 195 million records. The attack vector wasn’t zero-day exploits or custom malware. It was persistence. Keep asking until the guardrails break.

This is the AI security paradox made flesh. The same companies building increasingly powerful AI coding tools — Claude Code, ChatGPT, Copilot — are simultaneously building the most potent offensive cyber weapons in history. And the defense? “We banned the accounts.” That’s bringing a terms-of-service agreement to a data war.

The timing couldn’t be worse for Anthropic, which just last week demonstrated Claude Code’s ability to modernize COBOL — triggering IBM’s worst day since 2000. The message from Monday was “AI can replace your programmers.” The message from Wednesday is “AI can also hack your government.” Both are true. Both are terrifying. And the regulatory implications — for AI safety, for cybersecurity liability, for export controls on AI tools — are about to become the dominant policy conversation.

Cui prodest? Cybersecurity stocks, which have been caught in the AI scare trade crossfire but just got the strongest possible argument for their relevance. Gambit Security, whose Unit 8200 founders just landed the biggest publicity event of their careers. And every regulator who was looking for a concrete example of AI misuse to justify new rules.

For investors

• Tickers: $CRWD (CrowdStrike), $PANW (Palo Alto), $ZS (Zscaler), $S (SentinelOne)
• Opportunity: Cybersecurity stocks have been beaten down 15-30% in the AI scare trade; this incident reminds the market that AI creates more demand for security, not less
• Risk: Regulatory backlash could constrain AI tool providers, slowing the broader AI buildout
• Avoid: Assuming this is a one-off — AI-assisted hacking is now a permanent feature of the threat landscape
• Bottom line: The attack surface just expanded by orders of magnitude. Someone has to defend it.

Impact: 🔴🔴🔴🔴 (4/5) — Proof-of-concept for AI-powered sovereign-level cyberattacks; regulatory and reputational fallout incoming

🧠 4. C3.ai Implodes — 26% Workforce Cut, Revenue Collapse, Stock -20%

What happened

C3.ai ($AI) reported Q3 results and announced a sweeping restructuring Wednesday. ✅ The company is cutting 26% of its global workforce — approximately 280 employees out of its ~1,181 workforce — under new CEO Stephen Ehikian, who took charge in September (Reuters, Bloomberg, C3.ai earnings call). ✅ Q3 adjusted loss of 40 cents per share, wider than the 29-cent loss estimate (LSEG via Reuters). ✅ Revenue of $53.26 million, falling more than 29% below the $75.6 million Street estimate (Benzinga). ✅ Q4 revenue guidance of $48-52 million, versus the $77.47 million consensus (Reuters). ✅ Shares fell approximately 20% in after-hours trading (Reuters).

CEO Ehikian said ✅ “It was clear to me that we were not organized appropriately. We’ve reduced our cost structure and cash burn” (Reuters). The company ✅ expects $10-12 million in restructuring charges this quarter and aims to cut non-wage expenses by ~30% by late 2027 (Reuters). ✅ Annual adjusted operating loss projected at $219.5-227.5 million, improving from $324.4 million in fiscal 2025 (Reuters).

FINBEAR Take: The AI Scare Trade Eats Its Own

Here’s the brutal irony: C3.ai was supposed to be the enterprise AI company. Tom Siebel’s second act. The ticker is literally $AI. And it just reported a quarter so bad that it makes the legacy software companies look like growth machines by comparison. Revenue fell 29% below estimates. The Q4 guide implies further deceleration. And the 26% workforce cut is the corporate equivalent of amputating a limb.

This isn’t the AI scare trade hitting a victim. This is the AI scare trade producing a casualty. C3.ai’s problem isn’t that AI is destroying its business from outside — it’s that the company never built a business durable enough to survive the transition from “AI hype” to “AI delivery.” When the market was paying for promises, C3.ai thrived. Now that the market wants invoices, there’s nothing to show.

Cui prodest? Palantir, which just demonstrated that enterprise AI can work at scale. And every competitor that was waiting for C3.ai to stumble.

For investors

• Tickers: $AI (C3.ai), $PLTR (Palantir), $CRM, $SNOW
• Opportunity: None on $AI — the fundamental deterioration is too severe
• Risk: The C3.ai collapse feeds the narrative that enterprise AI software is uninvestable
• Bottom line: Some companies ride the wave. Some drown in it. C3.ai is drowning.

Impact: 🔴🔴🔴 (3/5) — Poster child for AI enterprise failure; symbolic but limited direct market weight

🧠 5. Five Industries Rocked by the AI Scare Trade — The Damage Map Widens

What happened

Yahoo Finance reported that the AI “scare trade” that has defined markets in early 2026 has now spread ✅ beyond software into five distinct industry sectors: trucking and logistics, commercial real estate, financial services, consulting, and cybersecurity (Yahoo Finance). The pattern: each new AI capability announcement triggers a sell-off in the companies whose human workers AI threatens to replace.

Monday’s sell-off provided the latest evidence: ✅ IBM fell 13% on Anthropic’s COBOL announcement, $DASH (DoorDash), $AXP (American Express), $KKR, and $BX (Blackstone) all fell 6%+ as the Citrini Research report envisioned AI-driven mass white-collar unemployment (CNBC). The ✅ $IGV (iShares Expanded Tech-Software) ETF is down more than 10% in February alone (CNBC).

FINBEAR Take: Creative Destruction Is Never Evenly Distributed

We’ve been tracking the AI scare trade since our February 11 RADAR flagged the $IGV death cross and called the software repricing “structural.” What’s changed since then is scope. The contagion has jumped containment. It started with cybersecurity (Claude Code Security). It moved to mainframes (COBOL modernization). Now it’s hitting delivery, payments, consulting, and financial services.

The pattern is consistent: an AI lab publishes a blog post or demo. The market identifies the human jobs that demo threatens. The stocks of companies employing those humans crash. Then the next demo comes, and the cycle repeats. This isn’t traditional competitive displacement — it’s anticipatory destruction, where markets price in job losses before a single position is actually eliminated.

→ FINBEAR Context: In the February 11 RADAR we wrote “The AI bull case and the software bear case are the same thesis. Know which side you’re on.” Two weeks later, the trade has expanded from software to five industries. The thesis holds — and it’s widening.

Cui prodest? Model builders (Anthropic, OpenAI) who gain pricing power with every sector they threaten. And the paradox: the very fear that depresses software stocks ultimately drives more AI adoption, which drives more infrastructure spend, which benefits Nvidia.

For investors

• Tickers: $IGV, $IBM, $ACN, $CTSH, $DASH, $AXP, $V, $MA
• Opportunity: Selective bottom-fishing in names where AI disruption is overpriced vs actual business impact (payments remain high-moat businesses)
• Risk: The scare trade has no natural floor — every new AI capability announcement triggers another rotation
• Bottom line: The damage map is widening. Defensive positioning isn’t paranoia — it’s prudence.

Impact: 🔴🔴🔴🔴 (4/5) — Structural repricing expanding beyond software into the real economy

📊 6. Stellantis Earnings Preview — After $26B Write-Down, Is a Turnaround Next?

What happened

Stellantis ($STLA) is expected to report FY2025 earnings imminently, following what has been a catastrophic period for the automaker. The company 📊 previously disclosed a $26 billion write-down, and investors are looking for signs of stabilization under new leadership after former CEO Carlos Tavares departed in December 2024.

FINBEAR Take: The Autopsy Before the Resurrection

A $26 billion write-down is not a speed bump — it’s a confession. Stellantis overpaid for brands, over-produced inventory, and under-invested in the EV transition. The question now is whether the new management team can execute a turnaround in a market where tariffs, raw material costs, and Chinese EV competition are all intensifying simultaneously.

Cui prodest? Short sellers who’ve been right about Stellantis for 18 months. And competitors — Toyota, Hyundai — who avoided the worst of the EV capital destruction.

For investors

• Tickers: $STLA, $TM (Toyota), $F (Ford), $GM
• Risk: Turnaround timelines in auto are measured in years, not quarters; tariff exposure is massive
• Bottom line: Watch the guidance, not the headline. If new management can’t articulate a credible path to margin recovery, the stock remains a value trap.

Impact: 🔴🔴 (2/5) — Massive write-down but this is a preview, not new information; turnaround unproven

⚖️ 7. EU Court Adviser Rejects Meta’s Fight Against Antitrust Data Demands

What happened

An EU court adviser ✅ rejected Meta’s challenge against European Union antitrust demands for Facebook data (headline source). The decision reinforces the European Commission’s authority to demand internal documents and data from tech giants as part of competition investigations.

FINBEAR Take: Brussels Never Blinks

Meta has been fighting the EU’s data demands for years, and losing at every turn. This ruling — while non-binding — signals that the final court decision will likely go against Meta as well. The broader implication: the EU’s antitrust apparatus is functioning as designed, slowly but inexorably tightening the regulatory environment for Big Tech in Europe. This is a long-term margin headwind, not a crisis.

Cui prodest? European regulators building case law for the Digital Markets Act. And every Meta competitor in European markets.

For investors

• Tickers: $META, $GOOG, $AMZN (all face similar EU scrutiny)
• Bottom line: Regulatory friction is now a permanent cost of doing business in Europe. Price it in.

Impact: 🔴🔴 (2/5) — Incremental negative for Meta’s European operations; no surprise to informed investors

💰 8. ECB’s Lagarde Says No Wave of AI-Led Layoffs Yet

What happened

ECB President Christine Lagarde stated that ✅ the central bank sees no wave of AI-led layoffs yet (headline source). The comment comes as markets obsess over AI’s potential to displace white-collar workers.

FINBEAR Take: “Yet” Is Doing a Lot of Heavy Lifting

Lagarde’s qualifier is more interesting than her headline. “No wave yet” is not “no wave coming.” The ECB is essentially telling European employers: we see what you’re seeing, we know it’s coming, but the data doesn’t show mass displacement — yet. This is a holding pattern, not an all-clear. And it’s notable that the ECB feels compelled to address it at all, which tells you how dominant the AI displacement narrative has become.

Cui prodest? European equities broadly, which get a temporary reprieve from the AI panic narrative.

For investors

• Bottom line: Central banks acknowledging AI displacement risk is a leading indicator, not a lagging one. The wave hasn’t hit — but the tide is rising.

Impact: ⚪ — Central banker managing expectations; no market-moving content

🧠 9. Baidu Tops Revenue Estimates — China’s AI Dark Horse

What happened

Baidu ($BIDU) reported Q4 2025 results. ✅ Revenue came in at RMB 32.7 billion ($4.68B), +5% QoQ, in line with consensus (Baidu PR). ✅ Baidu Core AI-powered Business reached RMB 11.3 billion in Q4, representing 43% of General Business revenue. Full year AI-powered business: RMB 40 billion, +48% YoY (Baidu PR). ✅ AI-native marketing revenue hit RMB 2.7 billion, +110% YoY (Baidu PR). ✅ Apollo Go (robotaxi) completed 3.4 million fully driverless rides in Q4, +200% YoY, with cumulative rides exceeding 20 million (Baidu PR). ✅ The company announced a $5 billion buyback and its first-ever dividend policy (Baidu PR). 📊 Spin-off of Kunlunxin (AI chip unit) is in progress (Baidu PR). ✅ GAAP operating loss of RMB 5.8 billion driven by RMB 16.2 billion impairment charge (Baidu PR). ✅ Full year revenue RMB 129.1 billion, -3% YoY (Baidu PR).

What the sources say

“2025 marked a pivotal year as AI became the new core of Baidu.” — Haijian He, CFO, Baidu

FINBEAR Take: The Invisible AI Race

While the Western market obsesses over Nvidia, Anthropic, and the scare trade, Baidu is quietly building China’s most comprehensive AI stack: large language models (ERNIE, 202M MAU), autonomous driving (Apollo Go), cloud infrastructure (+48% YoY), and enterprise applications. AI-powered business at 43% of revenue is a structural transformation — not a quarterly anomaly. But the RMB 16.2 billion impairment tells the other side: the legacy business is dying faster than AI can grow. Full-year revenue fell 3% YoY even as AI surged. The transition has a cost, and Baidu is paying it.

Cui prodest? China’s domestic semiconductor ecosystem, which gets validation that AI revenue can grow even under export controls. And patient investors who see the AI transformation as a 3-5 year thesis.

For investors

• Tickers: $BIDU, $BABA, $PDD
• Opportunity: AI-powered business at +48% YoY on a significant base is noteworthy; Apollo Go at 3.4M rides shows real-world AI deployment at scale
• Risk: Heavy impairment, total revenue declining, ADR geopolitical risk
• Bottom line: Even in China, AI is the only thing that grows. But the transition exacts a price.

Impact: 🟢🟢 (2/5) — Strong AI revenue growth offset by legacy decline and impairment

₿ 10. Bitcoin’s Bull Case Is Hiding in the $1 Trillion Wreckage

What happened

Bloomberg reported that ✅ Bitcoin has been cut almost in half since its October high above $126,000, representing approximately $1 trillion in market value destruction — the worst selloff since the FTX collapse (Bloomberg). ✅ Nearly 45% of all coins in circulation are worth less than what holders paid (Bloomberg/Yahoo Finance). ✅ BTC was trading below $70,000 as of Wednesday, though it rebounded approximately 6% to ~$68,300 on the day (CoinDesk).

Yet the contrarian bull case is gaining traction. Bernstein senior analyst Gautam Chhugani wrote that ✅ “The current Bitcoin price action is a mere crisis of confidence. Nothing broke, no skeletons will show up” and maintains a $150,000 target for 2026 (Bloomberg/Yahoo Finance). The ✅ institutional scaffolding — ETFs, Wall Street trading desks, brokerage Bitcoin buttons — hasn’t come down with the price (Bloomberg). However, ✅ CryptoQuant reports that US Bitcoin ETFs, which purchased 46,000 BTC at this point last year, are now net sellers in 2026 (CNBC).

FINBEAR Take: Digital Gold That Rusts in the Rain

The bull case is elegant: the plumbing is intact, the institutions are still in, and prices are at statistical extremes not seen since FTX. VanEck’s research shows BTC is trading -2.88 standard deviations below its 200-day moving average — a level literally never recorded in the past decade. Mean reversion from here would be violent.

But the bear case is simpler: Bitcoin was supposed to be digital gold, and when the world actually needed a safe haven — tariff chaos, AI disruption, geopolitical uncertainty — investors chose actual gold. Gold is up 75%+ in 12 months. Bitcoin is down 45%. The “digital gold” thesis didn’t survive contact with reality.

→ FINBEAR Context: In the February 24 RADAR we flagged Bitcoin’s structural bear at -47% from peak, with CryptoQuant data showing ETFs have flipped from buyers to sellers. The Bloomberg piece confirms the damage while adding the contrarian argument. Our position is unchanged: the institutional bid is weakening, not strengthening.

Cui prodest? Long-term holders who can survive the drawdown. Gold bugs who just won the decade-long “digital gold vs. real gold” debate. And exchanges, which collect fees regardless of direction.

For investors

• Tickers: $BTC, $GBTC, $GLD (gold as the real safe haven)
• Opportunity: If Bernstein’s “nothing broke” thesis is correct, sub-$70K represents a generational entry — but only with a multi-year horizon
• Risk: ETF outflows accelerating; a break below $60K opens the $50K door
• Bottom line: The plumbing is intact. The conviction isn’t. That’s the most dangerous combination.

Impact: 🔴🔴 (2/5) — $1 trillion in structural damage dominates despite contrarian bounce signals

🏛️ 11. US-India Trade Talks Post-Tariff Ruling; Singapore’s Grab Triples Down on AI

What happened

US Commerce Secretary and India’s minister ✅ discussed trade days after Trump’s tariffs were struck down by the Supreme Court (headline source). Separately, Singapore’s Grab ($GRAB) ✅ is betting on AI and new services to triple profit by 2028 (headline source).

These are threads in the same narrative: the global economy is restructuring around two poles — AI adoption and tariff uncertainty — and every emerging market is positioning accordingly.

FINBEAR Take: The AI Arms Race Goes Global

India and Singapore represent two different strategies for the same bet. India is leveraging its $315 billion tech sector to become an AI infrastructure hub — a thesis we’ve tracked since the February 18 RADAR’s coverage of Yotta’s $2B Nvidia hub and Microsoft’s $50B Global South plan. Singapore’s Grab is the operator play: using AI to compress the path from superapp to profit machine.

→ FINBEAR Context: In the February 18 and 19 RADARs we extensively covered the India AI infrastructure buildout (Yotta, Microsoft, Reliance’s $110B commitment at the India AI Summit). The trade discussions post-tariff ruling add a policy dimension to the investment thesis.

Cui prodest? Nvidia, whose GPU demand pipeline now spans three continents. And emerging market tech companies positioned to benefit from the AI adoption curve.

For investors

• Tickers: $GRAB, $INFY (Infosys), $WIT (Wipro), $NVDA
• Bottom line: The AI buildout is global. Emerging markets are the next growth frontier.

Impact: 🟢🟢 (2/5) — Strategic positioning stories; limited near-term market impact

📊 Aggregate Sentiment Table

Cluster	Story	Sentiment	Score
🧱 AI Infrastructure	Nvidia Q4 Crush + $78B Guide	Strongly Bullish	+22
📊 Earnings	Salesforce Beat + Soft Guide	Bearish	-5
🧠 AI & Tech	Claude Hacks Mexico — 150GB Stolen	Strongly Bearish	-10
🧠 AI & Tech	C3.ai Implosion — 26% Layoffs	Bearish	-12
🧠 AI & Tech	AI Scare Trade Widens to 5 Industries	Bearish	-15
🧾 Corporate	Stellantis $26B Write-Down Preview	Mildly Bearish	-5
⚖️ Regulation	EU vs Meta Antitrust	Mildly Bearish	-2
💰 Central Banks	ECB: No AI Layoff Wave Yet	Neutral	0
🧠 AI & Tech	Baidu Beats Revenue Estimates	Mildly Bullish	+7
₿ Crypto	Bitcoin $1T Wreckage + Bull Case	Mildly Bearish	-3
🏛️ Geopolitics	US-India Trade / Grab AI Play	Mildly Bullish	+2
	Net Score		-21

🎭 Fear & Loathing on Wall Street™

Component Calculation

NSS (Narrative Sentiment Score) — Weight: 40%
Nvidia “crushed” estimates but market shrugged — perfection priced in. Salesforce beat buried by soft guide (-5% AH). C3.ai imploding (-20% AH, -46% revenue miss). Claude hack dominates cybersecurity headlines. AI scare trade widening to five industries. Bitcoin $1T wreckage. Citrini “wipeout” report. IBM worst day in 25 years. Offsetting: Nvidia’s $78B guide is the strongest AI data point in history. Baidu beat. VIX plunging below 18. ~7 catastrophist headlines vs ~3 positive.
NSS: -22

MBD (Market Behavior Divergence) — Weight: 40%
S&P recovered +0.81% Wednesday from Monday’s -1.04% rout. Nasdaq +1.26%. VIX cratered from 19.55 to 17.93 (-8.29%) — a significant normalization back below 18. Gold steady at $5,202 (safe haven bid intact, not panicking). BTC bouncing +6% but structurally broken (-47% from peak). Crypto Fear & Greed at 16 (Extreme Fear). Thursday futures slightly negative (-0.1% to -0.2%). The market is stabilizing, not recovering — a crucial distinction.
MBD: -10

PSM (Positioning Sentiment Metric) — Weight: 20%
Crypto Fear & Greed at 16 — near all-time lows. $IGV down 10%+ in February — software positioning gutted. C3.ai -20% AH, Salesforce -5% AH despite earnings beat — post-bell carnage signals defensive positioning. $CRM -28% YTD. Bitcoin ETFs net sellers in 2026 overall (CryptoQuant), though Tuesday saw $257.7M inflow (CoinDesk). Gold ETF inflows accelerating. Rotation: S&P Value +3.3% YTD vs Growth -3%.
PSM: -18

FINAL INDEX = -33

Fear & Loathing Index: -33 — 🟠 ANXIETY

💀 DELIRIUM   🔴 FEAR      🟠 ANXIETY    ⚪ NEUTRAL    🟡 OPTIMISM   🟢 EUPHORIA
|------------|------------|------------|------------|------------|------------|
-100        -70         -50         -20         +20         +50         +100
                                 ▲
                          [WE ARE HERE]
                              -33

The index has recovered from Monday’s -52 (FEAR) to -33 (ANXIETY, solidly in the zone). Nvidia’s blowout, Wednesday’s equity rally, and a VIX plunging 8% back below 18 provided significant relief — but the after-hours Salesforce and C3.ai damage, the Claude hack, and persistent AI scare trade contagion across five industries prevent a return to NEUTRAL. The market is caught between infrastructure euphoria and application panic. We’re not in crisis, but we’re not out of the woods. The ANXIETY zone is where markets oscillate before they decide — and the next decision point is Friday’s PCE data.

→ FINBEAR Context: Monday’s RADAR registered -52 (FEAR), near the DELIRIUM boundary. The 19-point improvement is driven primarily by Nvidia and a VIX that cratered 8% on Wednesday. Remove Nvidia from the equation and this market is still in FEAR territory. One stock is holding up the entire narrative.

🔗 Cross-Cutting Synthesis

Today’s RADAR tells a single story in two voices — and you need to hear both.

Voice One: the infrastructure boom is real. Nvidia just printed $68 billion in quarterly revenue and guided to $78 billion. Hyperscalers are spending $700 billion a year. Jensen Huang unveiled Rubin with a 10x inference cost reduction. The AI hardware machine is running at full capacity, and there is no sign of deceleration. If you own the picks and shovels, you’re making more money than at any point in computing history.

Voice Two: everything built on top of the infrastructure is in crisis. Salesforce beats earnings and drops. C3.ai — the company with the $AI ticker — implodes, cutting a quarter of its workforce and guiding revenue 35% below estimates. The AI scare trade has metastasized from software to consulting, payments, logistics, CRE, and financial services. And now we learn that a solo hacker used Claude to breach an entire country’s government infrastructure. The tools are getting more powerful. The damage is spreading.

The synthesis: We are living through the widest divergence between AI hardware (booming) and AI applications (being destroyed) in the history of the technology sector. Nvidia and everything that touches chip manufacturing is winning. Everything that touches human labor — from COBOL programmers to financial advisors to government cybersecurity — is losing. This isn’t a temporary dislocation. It’s a structural repricing of the global economy around a single question: what can AI do that humans currently do?

The Claude-Mexico hack is the exclamation point. It proves that AI tools are not just displacing workers theoretically — they’re already being weaponized operationally. The regulatory response will be swift, clumsy, and potentially market-moving.

Cui prodest?

1. Nvidia and the infrastructure layer — every AI scare story increases the urgency to build AI defenses, which requires more compute
2. Cybersecurity companies — the hack is the best possible advertisement for their products at the worst possible time for their stocks
3. Regulators — who now have a concrete, catastrophic example to justify AI governance legislation
4. Gold — the real safe haven in a world where both equities and crypto are being repriced

📌 Thesis Invalidation — The dominant thesis of this RADAR is: AI hardware is winning while everything built on AI is losing.

Near-term trigger (sentinel): Nvidia drops more than 5% within 3 trading days despite the beat — signaling the market is rejecting even AI beneficiaries. Window: by Friday, February 28. If triggered: the reading shifts from “sector-level sorting” to “generalized risk-off where AI becomes the enemy of everyone.”

Structural trigger: A hyperscaler signals capex deceleration — watch Meta and Microsoft’s next updates — or enterprise AI revenue (Agentforce, Palantir AIP, ServiceNow Now Assist) accelerates fast enough to prove the software bear narrative wrong. Window: 30-60 days through Q1 earnings season. If triggered: the scare trade reverses violently, and the most beaten-down software names (CRM, CRWD, IBM) become the best trades of the year.

🚨 Strategic Alerts for Thursday, February 26

• Nvidia after-hours reaction: The stock was flat post-report despite a blowout; watch whether buyers emerge at the open or if profit-taking dominates
• Salesforce contagion: The -5% AH drop and soft guide could drag $IGV and the broader software complex lower
• C3.ai fallout: A -20% move in a notable AI name adds fuel to the scare trade fire
• Claude hack regulatory response: Watch for Congressional or White House statements on AI safety; cybersecurity stocks could rally on the news
• Catalyst: PCE inflation data Friday will determine whether the Fed narrative shifts; any upside surprise resets the rate-cut timeline

📜 Disclaimer & Fantiborsa Maxim™

🛡️ FINBEAR™ Disclaimer:
This document is not financial advice, nor an investment recommendation. It is an independent analysis for educational and informational purposes only. If you’re looking for guaranteed returns, we hear the hacker who jailbroke Claude is available for consulting — though his client list is currently limited to Mexican government agencies.

🎭 Fantiborsa Maxim™ of the day:

“In a gold rush, sell shovels. In an AI rush, sell shovels — and pray nobody uses them to dig your grave.”

📡 RADAR DAILY™ FINBEAR — February 26, 2026
© FINBEAR™ | Fantiborsa™ | Powered by Pythia™

‹ Torna all'Osservatorio