Oxx has published research on the backgrounds of founders behind successful software and AI companies, pointing to a marked shift in founder profiles between the SaaS era and the first wave of AI and machine learning businesses.
The investment firm analysed 645 founders across 270 companies: 150 business-to-business software-as-a-service groups that achieved exits or stock market listings above USD $500 million, and 120 AI and machine learning companies on what it described as a similar trajectory.
The findings suggest academic and technical credentials became more prominent as the market moved from SaaS to AI and machine learning. The share of founders with PhDs rose from 6% in the SaaS wave to 18% in the AI and machine learning wave, while the proportion who attended leading universities increased from 37% to 61%.
By contrast, MBA qualifications became less common in the later cohort, falling from 14% in the SaaS period to 4% in AI and machine learning.
The data also points to another shift within the current AI market. Among what Oxx described as the most recent generation of founders building post-2023 generative and agentic AI application-layer software companies, the indicative PhD rate fell to 12% from 18% in the earlier AI and machine learning group.
That suggests the educational profile associated with the earliest AI company builders may already be changing, as newer companies focus less on fundamental model development and more on software products built on top of existing AI systems.
Richard Anton, co-founder and general partner at Oxx and author of the white paper, said the earlier change from SaaS to AI reflected the demands of building more technically complex products.
"The increase in the rates of PhDs between the SaaS era and the beginning of the AI/ML era makes sense. In the SaaS era, commercial skills were necessary for scaling companies. In the early AI era, technical skills have been a bottleneck to B2B success - these products and technology are genuinely hard to build," Anton said.
Founder mix
Alongside formal education, Oxx found that having a technical co-founder remained one of the strongest common factors across both periods. It found that 85% of successful SaaS companies had a technical co-founder, compared with 95% of AI and machine learning companies.
That continuity stands out because other founder characteristics appear to have changed significantly over time. The report argues that while the specific technical skills required in software businesses have evolved, founding teams still tend to need a member with deep technical expertise.
Anton described that as one of the clearest patterns running through both generations of company building.
"One thing connects both generations is the technical expert. A technical co-founder was present on 85% of SaaS teams, and 95% of AI/ML. One thing remains constant: even if the nature of the technical skill might change, when it comes to co-founding teams, someone needs to be a technical expert," Anton said.
The report also links the rise in founders from leading universities to the networks and visibility those institutions can provide. Oxx highlighted MIT, Stanford, Oxford and Cambridge as examples of academic environments that featured more heavily in the AI and machine learning cohort.
Anton said attendance at those universities appeared to matter even when founders did not complete their courses.
"Not all the founders completed their degrees, but their attendance allowed them to build the right networks, build co-founder relationships and gain investor attention," Anton said.
Regional shift
The research also indicates that successful software company formation has become less concentrated in the United States. In Oxx's sample, the UK's share of successful companies rose from 1% in the SaaS wave to 12% in AI and machine learning, while continental Europe's share increased from 5% to 10%.
Those figures suggest a broader redistribution of software entrepreneurship towards European markets, particularly in AI-related businesses where university research strength and technical talent have become more important.
Anton said the UK's academic base had been an important part of that shift.
"The UK has done well because of its world-class AI teaching and research at its universities," Anton said.
He added that Europe's position had strengthened over several technology cycles as founder role models and funding became less exclusively American.
"I've been investing in software for 30 years and through various innovation cycles. I could always see that the UK and Europe had immense talent, but the role models and the capital back then were almost exclusively American. That has been eroded and the UK and continental Europe have proven they have the talent, institutions and market complexity to build global winners," Anton said.
Investor signals
For investors, the report argues that founder assessment patterns may need to change again as AI tools become easier to use. If the scarcity value of a PhD falls as the technology becomes more accessible, investors may place greater weight on a founder's judgement in identifying practical uses for AI and turning them into products.
Anton said that shift has implications for how venture capital firms read founder credentials.
"A large part of venture capital is pattern recognition - this will remain critical. However, investors need to recognise that the patterns worth recognising are changing, and which signals to look for, particularly when it comes to assessing founder talent," Anton said.