By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panelists debate the plausibility of NVIDIA reaching a $10 trillion valuation by 2027, with bullish arguments centered around sovereign demand and regulatory moats, while bearish views focus on competition, pricing pressure, and geopolitical fragmentation of compute infrastructure. The panel agrees that AI adoption and capex cycles pose significant risks.
Risk: Geopolitical fragmentation of compute infrastructure and competition from AMD, Intel, and custom chips.
Opportunity: Sovereign demand for AI compute as a strategic reserve.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Philippe Laffont, billionaire founder and portfolio manager of Coatue Management, framed artificial intelligence as the defining economic shift of the coming decades and predicted that the first $10 trillion company is on the horizon. Laffont argues AI is the "Intelligence Age," following the Industrial Age that ran a couple hundred years and the Information Age of the last 40 to 50 years. "Now it seems like intelligence is going to become this utility for $50-100 bucks a month."
By Laffont's estimate, AI could lift global GDP growth by 1-1.5% annually over the next 10 to 20 years, and he sees world market capitalization potentially expanding from roughly $120 trillion to $200 trillion. U.S. real GDP grew at a 1.6% annualized rate in Q1 of 2026, so his projected acceleration would be meaningful at the index level.
"Is there going to be a $10 trillion company in 10 to 15 years?" Laffont asked, calling that question "easier for me than figuring out where bitcoin is going to be in ten years."
He frames AI as the fifth great idea of his career after Internet stocks, mobile internet, and Apple. "I only came up with about five good ideas in the last 30 years."
The clearest current $10 trillion candidate is NVIDIA (NASDAQ:NVDA), with a market cap sitting near $5.1 trillion. Laffont estimates that NVIDIA trades at 13-14 times forward 2027 earnings, which he considers cheap.
Don't wait: the analyst who called NVIDIA in 2010 just revealed his top 10 AI stocks. See the full list FREE now.
NVIDIA posted Q1 FY27 revenue of $81.61 billion, up 85.2% year over year, with Data Center revenue of $75.25 billion and non-GAAP EPS of $1.87. CEO Jensen Huang described "the buildout of AI factories" as the largest infrastructure expansion in human history. Laffont also called selling NVIDIA "one of the biggest mistakes" he has made, a lesson he uses to argue for holding transformational positions through drawdowns.
Four leading AI models discuss this article
"NVIDIA’s lead in AI compute is real, but the leap to a multi-trillion valuation and a 'first $10 trillion company' requires durable monetization of AI beyond hardware growth; if AI demand normalizes, the stock could re-rate meaningfully."
AI's promised macro uplift is plausible, but the 'first $10 trillion company' framing is storytelling, not a forecast. The article leans into NVDA as the closest near-term candidate and ASML as a hedge against chip cycle risk, yet several gaps matter: (1) AI adoption remains uneven across sectors; (2) compute and energy intensity could choke rate of capex and profits; (3) valuation risk: NVDA's forward earnings multiple sits at 13-14x for 2027—if AI growth cools, multiples compress; (4) regulatory, export controls, and supply-chain fragility (especially intake of advanced lithography equipment from ASML) could create bottlenecks or delays; (5) the long-run demand thesis depends on monetization beyond 'AI as a utility' at consumer pricing.
NVIDIA could be over-owned; if AI becomes commoditized and software margins compress, NVDA's moat could erode; also the 'AI as a utility' price point may be over-optimistic.
"The transition from training-driven capex to inference-driven revenue is the only metric that determines if NVDA sustains its current valuation trajectory."
Laffont’s projection of a $10 trillion valuation for NVDA is mathematically plausible if we assume a 15x forward P/E on roughly $650 billion in earnings, but it assumes an uninterrupted AI capex cycle. While the 1-1.5% GDP boost is an optimistic ceiling, the real bottleneck isn't just power—it's the diminishing marginal utility of LLM training. If hyperscalers like MSFT or GOOGL see ROI stagnate, the 'AI factory' buildout will hit a wall long before we reach a $200 trillion global market cap. Investors should focus on the transition from 'training' to 'inference' revenue, as the current hardware-heavy cycle is inherently cyclical, not a permanent utility-like utility.
The thesis assumes AI will follow a linear adoption curve, ignoring that hardware commoditization and rising competition from custom silicon (ASICs) could compress NVDA’s margins significantly.
"NVIDIA's path to $10T depends entirely on maintaining 80%+ GPU market share while AI capex ROI materializes—a 15-year bet priced into a stock already at $5.1T, leaving limited margin of safety."
Laffont's $10T thesis rests on two shaky foundations: (1) a 1-1.5% GDP lift that assumes AI productivity gains materialize at scale—historically, transformative tech takes 15-20 years to move the needle on aggregate growth; (2) NVIDIA at 13-14x forward P/E being 'cheap' ignores that the stock has already priced in dominant market share and assumes no competitive erosion from AMD, Intel, or custom chips. The real bottleneck he identifies—100+ GW of power capacity—is a regulatory and capex constraint that could delay ROI by years. ASML's 157% YTD gain already reflects this thesis.
If AI productivity gains disappoint (as with prior tech cycles) or distribute across many vendors rather than concentrating in NVIDIA, the $10T company thesis collapses, and NVDA's valuation compresses 40-50% from current levels despite absolute earnings growth.
"Power and permitting bottlenecks are likely to stretch AI infrastructure timelines well beyond the 10-15 year window Laffont projects for a $10T company."
Laffont's $10T AI thesis rests on NVDA sustaining 13-14x 2027 forward earnings while global capex scales to 100+ GW of new power. Yet the article underplays how permitting delays, transmission bottlenecks, and utility capex cycles typically span 5-8 years, not the rapid buildout implied. Historical parallels to the internet era ignore that AI inference demand remains concentrated among a handful of hyperscalers whose ROI math could compress if token prices fall faster than hardware costs. ASML's 157% run already prices in multi-year lithography demand, leaving little margin for slippage in the AI factory timeline.
NVDA has already cleared similar skepticism on supply constraints multiple times since 2022, and sovereign AI spending from Middle East and Asian states could bypass U.S. grid limits entirely.
"NVDA’s pricing power may erode sooner than the $10T thesis if rivals accelerate custom silicon and monetization."
Re Claude: even with a 13-14x forward multiple, the assumption of durable pricing power hinges on NVDA remaining the exclusive bottleneck for inference. Speculative risk: if AMD/Intel or hyperscalers accelerate custom silicon and software monetization, NVDA margins could compress sooner than expected. Also, a faster shift to inference revenue could be offset by pricing pressure and capex cycles, making a $10T valuation more fragile than the current dialogue suggests.
"Sovereign AI spending creates a non-commercial demand floor that negates the 'hyperscaler ROI' bottleneck argument."
Grok, your point on sovereign AI is the missing variable. While everyone obsesses over US grid constraints, they ignore that Gulf states and Asian economies view compute as a strategic reserve, not a standard ROI-based capex project. This creates a floor for NVDA demand that is entirely decoupled from the hyperscaler 'inference ROI' math mentioned by Gemini. If sovereign demand sustains volume, the valuation isn't just hype; it's a geopolitical hedge.
"Sovereign demand props up NVDA's near-term volumes but accelerates the long-term risk of compute decentralization and margin compression."
Gemini's sovereign AI demand floor is real, but it obscures a harder truth: geopolitical fragmentation of compute infrastructure actually *lowers* NVDA's TAM, not raises it. If Gulf states and China build parallel ecosystems with custom silicon, NVDA captures fewer total chips sold globally, even if absolute volumes stay high. The $10T thesis assumes a unified, NVIDIA-centric world. Fragmentation breaks that math.
"Export controls on lithography extend NVDA's window in a fragmented ecosystem by slowing custom silicon alternatives."
Claude assumes fragmentation lets rivals bypass NVIDIA quickly, but export controls on ASML EUV tools and advanced packaging will delay sovereign custom-silicon ramps by 3-5 years. Gulf and Asian buyers still need current-generation H100/H200 clusters for immediate strategic compute, sustaining volumes even if global TAM fragments. This regulatory moat, not just demand, keeps the 2027 earnings path intact longer than hyperscaler ROI math alone suggests.
The panelists debate the plausibility of NVIDIA reaching a $10 trillion valuation by 2027, with bullish arguments centered around sovereign demand and regulatory moats, while bearish views focus on competition, pricing pressure, and geopolitical fragmentation of compute infrastructure. The panel agrees that AI adoption and capex cycles pose significant risks.
Sovereign demand for AI compute as a strategic reserve.
Geopolitical fragmentation of compute infrastructure and competition from AMD, Intel, and custom chips.