Later this month, we are bringing together a group of technology leaders from PE-backed portfolio companies and Technology Operating Partners from the funds themselves to discuss exactly this topic.
The reason is fairly simple. I spend a lot of time speaking with CTOs, Technology Operating Partners and investors across private equity, and one question is coming up more and more often: AI is clearly making software development faster, but is it actually making businesses better?
That distinction matters.
Most technology teams are now using some form of AI-assisted development, whether that is Copilot, Claude, Cursor or more agentic tooling. Developers can generate code faster, test faster and increasingly automate parts of the software development lifecycle.
But faster code generation does not automatically mean faster delivery, better quality or stronger R&D returns.
For PE investors in particular, the interesting question is no longer whether a portfolio company is “using AI”. Almost everyone is. The more important question is whether AI is genuinely changing the economics of the technology function and, ultimately, contributing to the investment thesis.
Access to AI will not be the differentiator
AI coding tools are becoming mainstream incredibly quickly.
GitHub has reported that almost 80% of new developers use Copilot within their first week, while Google’s DORA research found that around 90% of technology professionals surveyed were already using AI at work.
If adoption continues at that pace, access to the technology itself is unlikely to create much competitive advantage. Every CTO will be able to buy broadly the same tools and access broadly the same models.
The differentiation will come from the operating model around them.
How quickly can a business move from an idea into production? How strong is the architecture? How automated are testing and security? How good is the documentation? How much time is still being lost through reviews, hand-offs and internal decision making?
A business might make coding 30% faster, but if releases are still sitting in QA for two weeks, the business itself has not become 30% more productive.
That is the bit I think often gets overlooked.
The bottleneck is moving
AI does not necessarily remove bottlenecks from software development. It changes where they sit.
Historically, engineering capacity itself was often the constraint. There were only so many developers, so many hours in the day and only so much code a team could produce.
As code becomes easier and faster to generate, the constraints start moving into other parts of the system: architecture, product definition, testing, security, review, context and decision making.
Sonar’s recent research is a good example. Developers estimated that AI was already responsible for around 42% of committed code, yet 96% said they did not fully trust AI-generated code.
That creates an interesting tension.
Businesses may be able to create significantly more software, but that potentially places even more value on experienced engineers who can review it properly and understand what good looks like.
In other words, the scarce resource may increasingly shift from the ability to write code to the ability to judge it.
For PE-backed businesses, that is an important change.
PE should probably be asking different questions
I also think this changes how engineering productivity should be measured.
Lines of code have never been a particularly useful metric and they become even less useful in an AI-enabled environment. The same applies to the amount of AI-generated code, number of prompts or developer productivity viewed in isolation.
The more useful questions are commercial.
Can a portfolio company take a new product from concept to market in two months rather than six? Can technical debt be reduced without slowing the roadmap? Can newly acquired businesses be integrated faster? Can engineering teams support more products without headcount increasing at the same rate? Can new AI-enabled features create additional revenue or improve retention?
Those are the outcomes that start connecting engineering performance back to the value creation plan.
McKinsey recently looked at hundreds of PE-backed businesses and found that companies showing broader AI adoption traded at significantly higher revenue multiples than those using AI more selectively.
Clearly, that does not mean AI alone created the valuation difference. But it does reinforce something I am hearing much more often from investors: AI is moving from being an innovation topic into the investment case.
Technology leaders therefore need to be able to explain not just what they are implementing, but what value it creates.
The shape of engineering teams will change
There is an obvious conclusion that if AI makes engineers more productive, businesses will simply need fewer engineers.
I think the reality will be more nuanced.
Teams probably will get smaller in some areas, and Gartner is already predicting much broader adoption of smaller AI-enabled engineering teams over the coming years. However, that does not necessarily mean simply reducing headcount.
It is more likely to change the composition of teams and the skills businesses value most.
Strong platform engineers become more important because they create the environments, tooling and guardrails everyone else uses. Senior architects become more important because architecture becomes harder to protect when code generation accelerates. Product leaders become more important because building the wrong thing has suddenly become much easier.
The same applies to CTOs.
The best technology leaders in PE-backed businesses are already moving beyond being purely technology leaders. They understand R&D economics, they can explain where AI creates genuine leverage and they know when productivity improvements should translate into lower cost versus when additional capacity should be reinvested into growth.
Most importantly, they can connect technology decisions directly to the investment thesis.
That feels like a very different CTO brief to the one many businesses were hiring against five years ago.
This is becoming a PE capability as well
We are also seeing more sponsors build these capabilities centrally.
Vista has partnered with Google Cloud across its portfolio, Clearlake has announced partnerships around AI-enabled value creation, and Blackstone and Hellman & Friedman have backed Ode alongside Anthropic to help companies move AI initiatives from ideas into production.
That direction of travel makes sense.
Rather than asking every portfolio company to solve the same problems independently, funds can start creating repeatable frameworks around AI tooling, engineering standards, governance, data, measurement and leadership.
PE has already done this in areas such as pricing, procurement, sales effectiveness and finance transformation.
Technology and AI are increasingly becoming another portfolio-level value creation capability.
The diligence question is changing too
There is also a wider strategic point for investors.
If software becomes materially cheaper and quicker to build, diligence needs to look harder at what actually creates the moat in a software business.
Is it the technology itself, or is it the data, integrations, customer relationships, regulatory complexity, distribution and ownership of the underlying workflow?
That distinction becomes much more important if competitors can reproduce functionality significantly faster than they could before.
So the question during diligence is no longer simply, “How good is the technology?”
It is increasingly, “How defensible is this business if software becomes dramatically easier to build?”
And equally importantly: “Can this company use AI to move faster than the rest of its market?”
Where I think this lands
AI is clearly rewriting the SDLC, but I do not think the winners will simply be the organisations using the most AI or generating the most code.
They will be the businesses that redesign how software gets built around it: better platforms, better architecture, better testing, better measurement, better product thinking and better technology leadership.
For PE, that makes this much bigger than an engineering productivity conversation.
It becomes part of diligence, the value creation plan, organisational design, leadership hiring and ultimately the exit story.
Access to AI will become standard. What companies build around it is where the differentiation will sit.
And that is exactly why we are running this event later this month with Plandek: bringing together technology leaders from PE-backed portfolio companies and Technology Operating Partners from the funds to compare what is actually working, how productivity and R&D ROI should be measured, and what an AI-enabled software operating model really looks like in practice.
Get in touch
Tristan Dugmore leads our Technology Leadership, Data & AI practice, helping investors and CEOs build the leadership teams driving growth, transformation and AI adoption. If you are looking to hire, or find your next role, please contact Tristan on tristan.dugmore@phippscameron.com
