To The Moon

Here's a comprehensive, bullish counter-article based on the title "Burry's AI Obituary is Premature: Why Palantir and Nvidia Are Just Getting Started". It directly addresses Burry's key bearish points (AI bubble akin to dot-com, unsustainable hyperscaler capex, aggressive depreciation assumptions inflating earnings, lack of near-term ROI, and overvaluation of "poster children" like PLTR and NVDA) while highlighting why the long-term structural tailwinds make his shorts vulnerable—especially as we sit in April 2026 with mixed but resilient performance signals.Burry's AI Obituary is Premature: Why Palantir and Nvidia Are Just Getting StartedMichael Burry made headlines in late 2025 with massive put positions against Palantir (PLTR) and Nvidia (NVDA)—over $1 billion notional in puts, concentrated heavily on PLTR. He warned of an AI bubble propped up by questionable accounting (extended depreciation schedules on GPUs and infrastructure), hyperscaler capex that won't deliver ROI, and parallels to the dot-com era where spending peaked before reality hit. In early 2026, parts of his thesis appeared prescient: PLTR pulled back sharply (down ~35% from his entry point at times), and broader AI software names faced a "monetization fracture" with valuation compression.But calling the end of the AI era? That's the real premature obituary. Here's why Burry's bear case underestimates the durability, adoption curve, and economic transformation underway—and why both Palantir and Nvidia remain positioned for multi-year upside despite near-term volatility.1. Depreciation Debate: Smart Strategy, Not "Fraud"Burry's core critique targets hyperscalers (Microsoft, Meta, Google, Amazon, etc.) for stretching GPU/server depreciation from ~3-4 years to 5-6+ years, which he argues artificially boosts reported earnings by understating expenses. He sees this as masking weak ROI on massive AI builds, setting up a cliff when upgrades accelerate or spending discipline returns.Counterpoint: This isn't accounting sleight-of-hand—it's a rational response to real-world utilization and rapid but uneven tech evolution. Nvidia's latest chips (Blackwell ramping strongly in early 2026) deliver massive efficiency gains in inference and training, extending economic usefulness beyond pure obsolescence timelines. Hyperscalers aren't blindly optimistic; they're seeing accelerating demand for agentic AI, sovereign AI projects, enterprise copilots, and physical-world applications (robotics, autonomous systems).Nvidia's Q4 FY2026 results underscored this: $68.1B revenue (up 73% YoY), with data center exploding and guidance for $78B in Q1 FY2027. Supply commitments nearly doubled to $95B+, signaling customers locking in capacity through 2027—not pulling back. CEO Jensen Huang has repeatedly framed AI infrastructure as "the largest buildout in history," with hyperscaler capex projected near $700B combined in 2026. If ROI were illusory, we'd see cancellations or delays; instead, visibility into hundreds of billions in Blackwell/Rubin revenue persists.Extended lives make sense when utilization rates climb and newer architectures complement (not instantly replace) prior ones. Burry's 2-3 year economic life assumption feels overly aggressive given current deployment data.2. Capex "Unsustainability" vs. Structural Demand InflectionBurry draws dot-com parallels: capex-to-GDP spiking, circular deals (e.g., Nvidia/OpenAI ecosystem), and spending outpacing proven returns. He questions whether AI will ever justify the trillions in projected infrastructure.Reality check: Unlike the late-1990s internet (mostly speculative consumer apps with thin monetization), today's AI delivers measurable enterprise value today. Palantir's Artificial Intelligence Platform (AIP) is a prime example—bootstrapping ontology-based AI that turns messy data into deployable agents for defense, healthcare, manufacturing, and finance. Commercial remaining deal value surged, with U.S. commercial revenue guidance implying triple-digit growth potential in segments for 2026. Palantir forecasts ~$7.19B revenue (midpoint) for 2026, with strong margin expansion and free cash flow—hardly a bubble reliant on hype alone.Nvidia benefits directly: Inference demand is inflecting hard in 2026 (agentic systems, token generation economics), alongside training. Sovereign AI (nations building domestic stacks), AI-native startups, and industrial use cases add layers Burry's bubble narrative glosses over. Global AI factory buildout estimates reach $3-4T annually by 2029-2030. This isn't froth—it's infrastructure akin to electricity or cloud computing in prior decades: sticky, compounding, and defensive against recession (productivity gains offset costs).Short-term pullbacks (PLTR's recent 7%+ drops amid broader AI software skepticism) reflect digestion after 2025's euphoria, not the end. Nvidia's puts remain underwater for a reason—demand hasn't cliffed.3. Valuation and "Poster Children" Risks: Growth Justifies PremiumsBurry sees PLTR and NVDA as overvalued symbols of excess, with PLTR's high multiple vulnerable to any growth slowdown and NVDA pure-play risk if capex moderates.Yet fundamentals argue otherwise:Palantir: Profitability edge is widening. 2026 outlook shows robust revenue growth paired with operating leverage—adjusted EPS projections imply continued 50%+ expansion. Government contracts provide stability; commercial acceleration (AIP bootcamps converting to large deals) adds upside. Wall Street targets often sit well above recent levels, citing "supercycle" potential in enterprise AI.

Nvidia: 65%+ revenue growth in FY2026 from a massive base is rare. Gross margins hold ~75%, with roadmap clarity (Rubin in 2026, beyond). Chinese market nuances exist, but core U.S./global hyperscaler demand dominates. At current trajectories, AI's productivity multiplier could make today's multiples look reasonable in hindsight—much like cloud leaders post-2010.

Burry's history is legendary (2008 housing), but he's been early—or wrong—on tech shifts before. AI isn't a replay of dot-com pets.com; it's foundational tech with network effects, data moats, and defensibility (Palantir's ontology, Nvidia's CUDA ecosystem).The Bottom Line: AI Is Early, Not OverBurry's obituary reads like a victory lap ("shorts are not forever," with nods to triumphant themes), and PLTR's 2025-early 2026 volatility gave his puts gains. But Nvidia's resilience and Palantir's execution (sticky platforms, expanding margins) suggest the real story is acceleration, not exhaustion.We're in the infrastructure phase—painful capex now, transformative returns later. Hyperscalers aren't irrational; they're racing for competitive edges in an AI-native economy. Pullbacks create entry points for those betting on adoption curves, not timing peaks.Burry may profit on volatility, but the long thesis for Palantir and Nvidia isn't dead—it's just getting started. The companies "making all the money" in AI (as one CEO colorfully put it) have more runway than the bears admit.This is not financial advice. Markets are volatile; do your own research. AI stocks carry high risk due to execution, competition, and macro factors.