Pull up all insights from the Laffont brothers (Coatue)

The Laffont brothers characterize the current era as an AI "supercycle," a foundational technological wave poised to expand technology's share of global GDP beyond its current 15% [1, 6]. They argue this cycle is creating a new class of outperforming public companies and, more significantly, offers a powerful macroeconomic counter-narrative to fears of a U.S. "debt spiral" [1, 14]. The core of this thesis is that AI-driven productivity gains, potentially reaching **2.5-3.5% annually**, could enable the economy to grow its way out of its deficit, a scenario the bond market may already be pricing in [1, 17]. Early evidence of this leverage is seen in MAG7 companies, which have compounded revenue at over 20% in recent years while operating expenses grew at only 2% . Despite these productivity gains, Coatue's portfolio companies are not planning to cut engineering staff, suggesting a focus on reinvestment over cost reduction .

The battle for AI dominance is most visible at the infrastructure layer, where the allocation of scarce NVIDIA GPUs has become a primary leading indicator of future market leadership [7, 16]. Microsoft has been particularly aggressive in securing GPU supply, with its AI operations estimated to be producing 100 trillion tokens per month [1, 5]. In contrast, Amazon's position appears more tenuous; its **20% share** of NVIDIA GPU allocation is less than half its 44% cloud revenue market share, suggesting it is either behind in AI or pursuing a different, and perhaps riskier, hardware strategy [11, 18]. This infrastructure build-out is already causing market disruption, as proprietary data indicates that after a user subscribes to ChatGPT, their Google page views decline by 8% to 11% year-over-year [9, 10]. ChatGPT's user adoption and engagement have proven resilient despite significant competition, outpacing previous technology platforms [13, 15].

For private companies, this technological shift necessitates a clear-eyed strategy based on growth and profitability, as the one-size-fits-all approach of the ZIRP era is no longer viable . Coatue proposes a framework where profitable companies growing over 25% should prepare for an IPO, while those growing under 25% should reinvest in growth through M&A or new products [23, 30]. This disciplined approach is critical given the poor performance of recent market entrants; the **2021 IPO cohort** was down 50% five years after listing . Looking forward, a significant wave of AI-related IPOs from companies like SpaceX and Anthropic is expected within the next 12 to 24 months, with a predicted combined value greater than all unicorn exits over the past decade [27, 29]. This activity is supported by projections that the AI ecosystem's revenue could double to $600 billion by 2027 . Beyond equity markets, the Laffonts also foresee innovation in debt, predicting that the U.S. government may soon issue tokenized bonds directly to consumers [2, 22].