Zeki Insight: The goal of hiring is not to identify the most prestigious candidate. It is to identify the most capable one. The further a process moves away from measurable evidence and towards proxies, the greater the risk that valuable talent is overlooked.
Deep tech likes to think of itself as meritocratic.
In an industry built on mathematics, engineering and scientific discovery, the assumption is that the best ideas win and the best people rise to the top. Yet one of the most persistent biases in deep tech hiring is surprisingly old-fashioned.
Pedigree. The university on a CV. The laboratory where someone trained. The famous professor they worked under. The logo of a previous employer. In a field obsessed with technical precision, we still rely heavily on signals that are easy to see rather than signals that are most likely to predict future performance. The uncomfortable reality is that many companies building frontier technology still evaluate talent using signals that are only loosely connected to future performance.
That matters because the competition for exceptional technical and commercial talent has never been more intense. The Stanford AI Index 2026 paints a picture of an industry accelerating at extraordinary speed. Frontier model performance continues to improve. AI adoption continues to rise. Competition between countries and companies is intensifying. Underneath all of these trends sits a less-discussed reality: a relatively small number of highly capable people are shaping a disproportionate share of AI’s progress.
The challenge for employers is that exceptional people do not always look the way hiring systems expect them to.
Prestige Is Not The Same Thing As Capability
Elite institutions matter. They attract brilliant students. They conduct important research. They produce many of the world’s most successful scientists and engineers. But there is a difference between acknowledging that elite institutions contain talent and assuming that talent only comes from elite institutions.
University of Colorado researchers published in Nature found an extraordinary concentration in academic hiring: just 20.4% of U.S. universities produced 80% of tenure-track faculty. The researchers described prestige as a central organising force in academic hiring networks, creating highly persistent hierarchies over time. This matters for deep tech because so much technical hiring inherits academic assumptions. The effect extends far beyond academia.
Many deep tech hiring processes continue to inherit these assumptions. Prestige becomes a proxy for potential. Educational pedigree becomes a proxy for competence. Association becomes a substitute for evidence.
Proximity to excellence is not the same thing as excellence itself.
The person who worked alongside great researchers is not necessarily the person driving breakthroughs. The person who attended the most prestigious university is not necessarily the person best equipped to solve tomorrow’s problems.
The Frontier Economy Runs On Proven Talent
One of the most important insights to emerge from recent AI markets is that talent itself has become a strategic asset. Across the industry, companies are spending billions to secure teams, researchers and operators with highly specialised expertise. The market is increasingly valuing people in the same way previous generations valued intellectual property or infrastructure. This changes the hiring question.
Instead of asking where somebody studied, organisations need to ask what evidence exists that they create value. The most useful signals are often hiding in plain sight. Not just school names. Not just marquee employer logos.
Patterns. When you study innovators, researchers, engineers and GTM leaders over long periods of time, certain signals appear repeatedly.
- Demonstrated output. In technical roles, this may include publications, patents, inventions, open-source contributions, breakthrough research, or evidence of solving difficult scientific problems. The relevant question is not where someone learned their skills, but whether they have repeatedly converted expertise into meaningful outcomes.
- Depth of specialisation. Frontier industries increasingly reward people who develop expertise in narrow, difficult domains. AI, quantum computing, robotics, semiconductors and advanced biotechnology all require accumulated knowledge that takes years to build. What matters is not the brand attached to the individual but the substance of their expertise.
- Peer recognition. The best people are often recognised by other exceptional people long before they become visible to the wider market. The teams someone joins, the collaborators they attract and the networks they move within can provide stronger evidence than conventional credentials.
- Career trajectory. Progression is rarely random. People who repeatedly earn responsibility, move into increasingly complex challenges or become sought after across multiple organisations often reveal something more valuable than pedigree: sustained performance.
For GTM leaders, the equivalent evidence may be category creation, repeated market expansion, customer adoption, hiring high-performing teams, or the ability to build commercial systems from scratch.
When Proxies Replace Evidence, Bias Follows
The challenge with pedigree-based hiring is not simply that it favours certain universities or employers. It is that once a hiring process becomes reliant on proxies, it rarely stops there. The same tendency that encourages recruiters to favour a handful of institutions can influence how they evaluate gender, nationality, geography, accents, career paths, or visas. In each case, visible characteristics can become stand-ins for judgments about capability.
The question is whether employers are evaluating the best available evidence. And in deep tech, that distinction matters enormously.
Talent Is More Globally Distributed Than Hiring Assumptions Suggest
One of the clearest themes emerging from the Stanford AI Index is the increasingly international nature of advanced AI talent.
The report highlights a global race for AI capability, with research excellence, patents, publications and frontier model development emerging from multiple countries rather than a handful of traditional centres. The performance gap between leading U.S. and Chinese AI models has narrowed dramatically, while countries including South Korea have established leadership positions in areas such as AI patent density.
For employers, that creates an important challenge.
The most valuable engineer for a role may not be located in Silicon Valley. The most important researcher in a niche domain may not have spent their career at a globally recognised institution. The next exceptional GTM hire may come from a market that most recruiters rarely search.
Evidence increasingly suggests that innovation capability is geographically distributed, even if visibility is not. This is one reason why talent flows have become such an important signal. As leading researchers, inventors and operators move between organisations and countries, they reveal where expertise is accumulating and where new centres of excellence are emerging.
When Many Small Decisions Add Up To Bias
The challenge is often not overt discrimination. It is the accumulation of small decisions driven by familiarity, networks and pattern matching.
Start-ups frequently hire through trusted introductions, founder networks, previous colleagues and investor referrals. These channels are efficient, but they also tend to reproduce existing demographics. Teams built through narrow networks often end up hiring people who look similar, think similarly, and have followed similar career paths.
The same network effects that influence investment decisions often influence hiring decisions. In both cases, access to opportunity can be shaped by who is visible, known and trusted within existing ecosystems. One reason this matters is that underrepresentation alone does not prove a lack of capability.
Across the start-up ecosystem, women remain substantially underrepresented among founders, technical leaders and venture-backed company executives. Yet funding outcomes suggest that access to opportunity remains highly concentrated.
McKinsey noted that women-founded teams received just 1.9% of U.S. venture capital funding in 2022, raising longstanding questions about whether funding decisions consistently capture the full available pool of entrepreneurial talent. PitchBook data highlighted by the World Economic Forum found that women-only founding teams continued to receive around 2% or less of venture capital funding in both Europe and the United States.
These figures do not, by themselves, prove hiring discrimination. They do raise a different question:
If access to opportunity, investment and leadership remains so concentrated, how much high-value talent is being missed because decision-makers continue to rely on familiar signals?
This matters particularly in deep tech because the industry’s talent constraints are already severe. The Stanford AI Index 2026 shows that frontier AI capability is increasingly concentrated around a relatively small pool of highly specialised talent. Companies cannot afford to reduce the available talent pool further by relying on imperfect proxies.
When founders hire people who remind them of previous successful employees, previous colleagues, or themselves, they may believe they are identifying quality.
Sometimes they are simply identifying familiarity.
The stronger approach is to evaluate the evidence directly:
- What has this person built?
- What difficult problems have they solved?
- What expertise have they accumulated?
- What outputs have they produced?
- Which high-performing teams continue to seek them out?
- What impact have they demonstrated across different environments?
Because the goal is not to build teams that look familiar. It is to find people capable of producing exceptional outcomes.