That's dubious, and why many people have been calling on the UK to expose their analysis to peer review.
Since early contagion is an exponential process the rate of growth (and all higher derivatives) also grows exponentially along with the number of infected. As a result, "as early as possible" also ends up being the correct time to maximally spread out the peak with the least intrusive countermeasures.
In order to conclude that delaying mitigations will do a better job of reducing the peak they must be making some highly novel assumptions. ... or just not actually reasoning about it formally at all and instead depending on intuition which is often not particularly accurate for exponential processes.
Since early contagion is an exponential process the rate of growth (and all higher derivatives) also grows exponentially along with the number of infected. As a result, "as early as possible" also ends up being the correct time to maximally spread out the peak with the least intrusive countermeasures.
In order to conclude that delaying mitigations will do a better job of reducing the peak they must be making some highly novel assumptions. ... or just not actually reasoning about it formally at all and instead depending on intuition which is often not particularly accurate for exponential processes.