Interview
a16z Podcast | From Research to Startup, There and Back Again
- Fundamental hardware breakthroughs are expected to take a long time to emerge due to the slow hardware sector pace, with global adoption driven by a mobile and IoT explosion, while the total number of processor chips worldwide (including embedded systems) has reached approximately 50 billion.
- As the industry reaches the end of Moore's Law, energy efficiency is predicted to become the primary concern and the second biggest cost in large data centers after the physical cost of servers, causing CISC architectures to fall far behind RISC in battery-powered and data center applications.
- RISC architecture is forecast to widen its dominance gap over CISC as energy efficiency becomes paramount, a trend that will require further architectural shifts in both power and performance.
- Following early 1980s invention and late 1980s market attempts, RISC architecture is expected to take until the mid-1990s to achieve market dominance.
- The iPhone launch in 2007 is identified as the definitive takeoff point for the smartphone market, distinguishing it from earlier devices like Nokia phones which were not full computers.
- Software products are expected to transfer more easily from university to industry than hardware products, as graduate students are predicted to be incredible programmers capable of building impressive software shortly after graduation, though often not yet scaled for millions of users.
- Companies with paying customers are expected to be forced to reengineer products to meet commercial standards, a requirement distinct from the academic setting.
- Higher prices are proposed as a rule for success because they force customers to commit and take ownership of the technology.
- During the 2008 financial crisis, Stanford lost 28 percent of its endowment within a six-month period, necessitating immediate large-scale restructuring to avoid five to ten years of small budget cuts.
- The startup ecosystem is predicted to continue relying on a very slow deliberative process within complex organizations like universities, contrasting sharply with the speed required in startups.
- Investors and founders require full commitment from co-founders, creating a potential mismatch when professors attempt to join startups part-time while maintaining academic roles.
- Prioritizing technology transfer over fee extraction is expected to result in better researchers and teachers who return with wider experience.
- There is a risk that faculty leaving for the AI industry could hurt the long-term industry by "eating the seed corn" of academic research.
- China is predicted to develop a Silicon Valley due to entrepreneurial people and risk capital, though issues with liquidity and exits may require time to resolve.
- The United States may lay the foundation for a new Silicon Valley area if the current region becomes strangled by housing, traffic, and high living costs.
- Younger generations are expected to shift away from suburban dreams toward co-housing arrangements with 10 to 15 people to accommodate working 60 to 80 hours per week.
- The unique network effect created by management talent depth in the Valley, developed by companies like HP and Sun, is expected to prevent other regions from easily replicating the ecosystem.
- Humility in leadership is predicted to remove barriers to admitting mistakes and asking for help, provided it is balanced with decisiveness to maintain staff faith.
- The university system is expected to need educational changes to leverage technology to bend the cost curve; otherwise, student debt is predicted to continue rising significantly.
- A shift away from the traditional four-year degree toward a certification model based on mastery of specific course sequences is expected, particularly for post-baccalaureate education.
- A new breed of innovators filling the interstitial space between domain expertise and data science is predicted to become key as machine learning applies to biology, chemistry, and astrophysics.
- There is a risk that many entrepreneurs viewing AI/ML as the end of theory may produce "garbage in, garbage out" results by attempting to find value without understanding the domain.
- The academic system is shifting toward short-termism in incremental publishing, as industry players like Google and Microsoft now contribute more fundamental research due to superior resources, data, and computational power.
- The era of corporate research labs is predicted to be effectively over, having been replaced by an explosion of R&D at a far greater scale across the corporate landscape, as the previous model relied on pre-competitive capital offshoots of monopolies before 1975–1980.
- Self-driving car technology hit a tipping point with the winning of the DARPA Grand Challenge, marking a transition from academic experimentation to industrial application by companies like Google/Waymo.
- The number of young women entering computer science is undergoing a resurgence following a previous decline, driven by support groups reaching a critical mass to eliminate isolation.
- Computer science is expected to attract the very best students in many fields, surpassing biological sciences, with freshmen possessing more mathematics knowledge than previous generations' seniors.
- Engineers are predicted to be significantly more productive in software development today due to sophisticated tools, infrastructure, and instant access to online resources like Stack Overflow.
- The computer science industry is expected to have professionalized such that many enter the field to make money rather than for the love of it, creating a more mercenary but framed workforce.
- A growing wave of talent native in both art/humanities and code is anticipated, merging disciplines that were previously kept separate.