Interview, Fireside Chat, Podcast
Citrini Research Breakdown: Agents, "Ghost GDP", Consumer Spend | Figma Earnings Beat
- Anticipates aggressive military and commercial deployment of Anthropic capabilities against China, alongside a rapid acceleration in AI capability improvements where weekly advancements outpace organizational adaptation.
- Forecasts a market shift where superior AI software renders current B2B offerings inferior, potentially causing "ghost GDP" disconnects due to productivity gains accruing to few while consumer income lags, leading to wealth concentration and potential consumer spend softening.
- Predicts significant value erosion or "maiming" for legacy public companies like Shopify, HubSpot, and Monday if agentic layers capture their value, contrasting with the durability of payment-focused entities like Toast and the accelerating growth of Palantir.
- Expects a "shrinking circle" for platforms like Figma, Replit, and Lovable as Claude potentially hosts entire websites and builds databases, threatening their protectable territory unless they adapt within the next 8 to 18 months.
- Projects that public markets may overreact to existing security features or "perfectly priced" momentum stocks trading at 40-50 times revenues, triggering corrections in entities like CrowdStrike while Atlassian represents a dislocation with beaten-down prices despite revenue acceleration.
- Envisions a future intelligence-led landscape where only applications infused with intelligence sell and grow quickly over the next decade, while non-intelligence apps face flat growth or decline.
- Notes that while 1-1.5% of US software jobs may be displaced, this is less significant than historical industry losses, though levered SaaS firms face a "long five-year grind" of expense cuts and headcount reductions to manage debt.
- Foresees the creation of 50-200 million revenue "Frankenstein B2B companies" via nominal-priced mergers that may IPO in 2027, driven by startups founded around 2021-2022 leveraging foundation models rather than foundation models capturing all revenue.
- Estimates OpenAI requires $110 billion in additional capital to reach a $280 billion revenue forecast by 2030, predicting it will remain the consumer mindshare winner while ruthlessly abandoning non-performing hardware or other plan segments.
- Identifies security, complex coordination, and fintech banking relationships as areas where AI disruption remains limited, while hyper-niche agents continue to succeed despite difficulties in building specific enterprise agents for giants like Shopify.
- Suggests that buying a basket of stocks at an average of three times revenues and eight times EBITDA will likely result in solid returns, whereas buying at higher multiples carries blowup risks if perfection is not maintained.
- Expects a two-year adoption cycle for rolling over all software from ServiceNow to DoorDash is highly unlikely due to consumer inertia regarding delegated decisions like food ordering, despite CTO optimism.
- Predicts that 100,000 decamillionaires will emerge from AI leaders, while many companies may go out of business or become irrelevant within two years if AI-driven growth stagnation impacts them, even if they retain customers.
- Anticipates that Venture Capital will remain a niche industry where success depends on partnership structures leveraging specific strengths, symbolized by moves like Jack Altman joining Benchmark, rather than solo practices.
- Forecasts that Anthropic will expand aggressively into security without replacing CrowdStrike's business entirely, as security layers diffuse through enterprises, while Cloud Code already surpasses mediocre engineering boards in audit and penetration testing capabilities.
- Notes that intelligence will infuse all software over the next decade, creating a mediation layer as the primary opportunity since enterprises will not build directly on Claude nor will Claude build all focus systems alone.
- Highlights that only Palantir currently possesses a competitive agent among public B2B companies, with revenue acceleration from AI agents unobserved elsewhere, while private companies are exploding in revenue.
- Predicts that high-priced stocks may suffer corrections if they were "priced for perfection," and that the market will see a contrast between winners like Shopify and losers like Klaviyo or Atlassian during a "great dislocation."
- Expects momentum investing to remain effective for 6-18 months but warns of potential failures for high-multiple stocks facing a risk of blowup if their perfect pricing is disturbed.
- Suggests that enterprise leaders will be "killed" by startups building necessary agents, while current agent capabilities in public companies are "lip service" due to the immense work required to train, onboard, and cleanse data, a bottleneck expected to resolve in two years.
- Predicts that AI-generated content like short-form stories will disrupt Netflix, while the "ghost GDP" concept where productivity gains do not scale with consumer income could lead to market structural dislocation or short-term recession before long-term GDP growth.
- Expects that Figma's Q4 2025 earnings will show accelerating growth and retention as "winning looks," though Claude Code may soon create designs as elegant as human designers, posing a threat to the platform's territory.
- Notes that most CEOs are stressed about the future, and while AI will boost productivity and corporate profits, the lack of great agentic use cases in public providers is due to current technical constraints.
- Anticipates that if a company builds an extremely high-value agent previously impossible, it can close millions in revenue in its first week, contrasting with the difficulty of building agents for large enterprises.