Conference Presentation, Panel, Fireside Chat, Interview
10 Years Too Early: On a Tech Career, Business, and Investing
- Early attempts to digitize services like streaming, plumbing, and photography were historically premature but are expected to succeed as the internet fully transforms the economy over a generational timeline, with the speaker asserting that purchasing "everything on the internet" is inevitable.
- By 2018, the internet is projected to have undergone a transformation comparable to the invention of TCP/IP, HTTP, and FTP, driven by the current early-stage platform shift in cryptocurrency and raw protocols.
- Artificial intelligence is predicted to become a ubiquitous, mandatory component of all software within 18 to 24 months, eliminating distinct AI startups as a category, while AutoML reduces neural network architecture as a differentiator, leading investors to favor vertical companies over horizontal platforms controlled by major platform holders.
- A sophisticated software tool chain, including labeling tools approaching the complexity of Photoshop, is expected to emerge over the next five to ten years, marking the first major break in development since von Neumann machines.
- Three major platform shifts—on-premise to cloud, mobile-first, and AI-first—are forecasted to complete within nine years of the firm's founding, with cryptocurrency identified as the current very early phase of a new generation internet.
- Future mobility is anticipated to rely on autonomous vehicles, including self-driving cars, buses, and scooters, while consumer value is expected to be generated primarily by AR in the long term and VR in the short term.
- The speaker predicts limited crossover success between Amazon and Alibaba in each other's regions, while noting that China's rapid growth in STEM and work ethic (characterized by "997" hours) may outpace Silicon Valley.
- Durable companies are advised to embrace a higher ratio of failures to successes, contrasting this approach with current trends of propping up ventures in China's real estate and venture sectors.
- Successful enterprise software models are expected to utilize open-source components for developer adoption and marketing rather than revenue, with paid versions achieving 90% gross margins through enterprise utilities, a strategy exemplified by GitHub's bottom-up approach.
- Organizations must actively manage structural changes when scaling from 15 to 75 and then to 100 employees to avoid internal dysfunction, while founders are encouraged to accept long time horizons and experimentation, potentially facilitated by acquisitions.
- Consumer applications competing with established giants like Facebook or Airbnb are expected to become viable in the crypto space within one to two years, contingent upon the prior establishment of foundational protocols for DNS, identity, naming, addressing, and distributed computation.
- A future crypto scenario is hypothesized where protocol designers capture 90% of the value via mechanisms like ICOs, inverting the traditional internet model where protocol creators received no direct compensation.
- A specific prediction notes that "nobody is going to be looking for AI startups" as a distinct category within 18 months relative to the statement's timeframe because AI will become embedded in all software.