Episode Summary
Executive Summary: The discussion rejects claims that “software is dead,” arguing instead that AI is redistributing value across the stack. Mike Cannon-Brookes frames AI as a productivity and product-creation boom that will shrink seats in many non-engineering functions, while expanding software spend overall through new use cases, services, and infrastructure. The conversation centers on TAM expansion, model-provider economics, public vs private market dynamics, and the rise of AI-native categories like legal and support.
Main Topics: Software is not dead; AI is reshaping it (Priority: 5/5): The panel argues that software remains alive, but winners and losers will shift. AI changes workflows, accelerates product creation, and pressures mature vendors that fail to re-accelerate. TAM expansion vs zero-sum thinking (Priority: 5/5): A major debate is whether AI spend simply cannibalizes existing software budgets or expands total spend by creating more productivity, more products, and more infrastructure demand. Atlassian’s AI strategy and business health (Priority: 5/5): Mike explains Atlassian is using multiple models, building an AI substrate, lowering COGS, and seeing accelerated cloud and RPO growth, positioning the company as an “above the fold” winner. Category rotation: support, legal, and services (Priority: 4/5): Participants discuss how AI is rapidly transforming customer support, legal, and consulting workflows, with some categories likely to shrink headcount while others gain efficiency and new demand. Harvey, legal AI, and valuation discipline (Priority: 4/5): The group uses Harvey as a case study for AI-native software valuation, debating whether its rapid growth justifies its high revenue multiple and what that says about TAM discovery. Public vs private company constraints (Priority: 3/5): Public SaaS firms face EPS and disclosure pressure, while private AI companies can spend aggressively. The conversation explores whether that creates an unfair competitive dynamic or just different capital-allocation rules. Super Bowl ads and signaling in AI (Priority: 3/5): The Anthropic/OpenAI ad controversy is framed as a sign of the times: expensive signaling, brand competition, and ego-driven marketing in a capital-rich AI boom.
Key Arguments: Software is not dead; the technology industry always experiences churn, with old competitors disappearing and new ones emerging. AI spend is not purely zero-sum because it can increase productivity, create more software, and shift spend into chips, cloud, consulting, and services. Model-provider revenue should be viewed as part of a broader revenue stack; spending on Anthropic can also drive AWS revenue and other infrastructure layers. Atlassian’s core areas—product and engineering, IT service management, HR service management, and customer service—are benefiting from AI rather than being destroyed by it. Non-engineering categories are at existential risk of seat reduction, while engineering/product remain an “island of stability” because AI enables more software creation. Harvey’s growth is impressive enough to suggest genuine TAM creation, but valuation discipline still matters because investors must compare current multiples against future growth requirements. Support is a particularly strong AI use case because it is input-constrained, text-heavy, and increasingly tied to action-taking automation, not just question answering. Public companies can still compete in AI if they invest aggressively in R&D and communicate the long-term story, despite short-term EPS pressure. The current market is highly consensus-driven: investors prefer backing the clear winner at a high price rather than taking early contrarian risk. Super Bowl ads from AI companies are more about positioning and signaling than efficiency, though they also reflect a broader willingness to spend in a hot market.
Data Points: Anthropic optimistic ARR forecast for 2029: $149 billion - Cited as a scenario that would consume a very large share of global software spend if realized. OpenAI estimated 2029 ARR: $180 billion - Used alongside Anthropic’s forecast to illustrate potential model-provider scale. Total worldwide software market: $700 billion - Referenced to question whether top AI model revenue would be additive or cannibalistic. Consulting/services spend: ~$1 trillion - Mentioned as a large adjacent budget that AI software could potentially capture. Atlassian cloud revenue growth: 26% quarter growth rate - Mike cited accelerating cloud revenue growth as evidence of continued demand. Atlassian cloud revenue scale: Well north of $5-6 billion - Used to show the size of the cloud business while still growing fast. Atlassian RPO growth: 44% - Presented as a key sign of accelerating customer commitment. Atlassian overall company growth: 23% - Referenced while discussing the company’s strong quarter. Atlassian service collection business: Largest at-scale business; growing very fast - Used to argue that service management is expanding rather than shrinking. Product and engineering share of Atlassian business: 40% - Illustrates how important engineering/product tools remain to the company. Atlassian R&D headcount: 10,000 people - Used to emphasize scale and the impact of LLMs on software creation. Replit revenue/headcount example: $300 million revenue, 300 people - Illustrated lean go-to-market structures in AI-era software companies. Harvey current ARR: $190 million - Used as the base revenue level for valuation and TAM discussion. Harvey valuation: $11 billion - Referenced as a high-multiple private-market AI legal bet. Harvey growth trajectory: $200 million to $600 million in one year (projected) - Used to show spectacular growth and potential TAM creation. Lawyer market spend: $200 billion per year - Estimated annual U.S. spending on non-partner lawyers, associates, and paralegals. Harvey active users: 100,000 - Used to derive implied revenue per user and assess market size. Harvey revenue per active user: $2,000 per user - Compared with incumbent legal software economics to assess expansion potential. Support category share of business: 3% to 7% - Estimated range for service/support as a share of every business. AI support revenue at Zendesk: $300 million+ - Mentioned as a sign of legacy business plus AI growth in support software. New AI companies raised: 14 companies in the last two years over $100 million - Used to show how crowded and well-funded the support/agentic landscape is. New unicorn up-round statistic: 40% of newly minted unicorns in Q1 last year had one or more up rounds by Q4 - Illustrates consensus capital piling into perceived winners. Super Bowl ad pricing example: $5 million to $8 million range - Referenced to discuss how AI companies are using premium ad inventory for signaling.
Pivotal Quotes: "The idea that software as a category is dead is ludicrous to me." — Mike Cannon-Brookes: Core rebuttal to the thesis that AI will eliminate software as an industry. "I think we just have to let the revenue show us the path to TAM." — Participant discussion: Captures the argument that market size should be inferred from actual growth, not assumed up front. "I think every category that I know of outside of engineering and product is at existential risk of shrinking seats." — Mike Cannon-Brookes: Summarizes the view that AI will reduce headcount in many non-build functions while accelerating software creation.
Implications: AI is likely to expand software demand overall but redistribute profit pools toward winners, infrastructure, and AI-native applications. Investors should focus less on static TAM and more on actual growth, durability, and where value accrues in the stack.