Short answer:
Some Romance languages originally developed in quite isolated geographical areas, Sardinia and central-southern Italian Peninsula, have preserved original Latin geminated consonants. In addition, such languages had a remarkable tendency to create new geminated consonants by the mechanism of assimilation.
Italian linguistic Matteo Bartoli formulated some principles, called the "Bartoli area norms", which can help us to interpret the geographical distribution of linguistic phenomena to infer some conclusions in relation to relative chronology of the different linguistic forms present in different Romance areas. The first one states that an area that is geographically more isolated than other ones tend to be more conservative, that is, usually preserves some features of an early stage of Latin language.
Latin had several geminated consonants: [p:], [t:], [k:], [f:], [s:], [m:], [n:] and [l:]. Sardinia, Tuscany and at least part of central-southern Italian Peninsula had a quite isolated situation with respect to the main communication routes in Late Antiquity and in the Early Middle Ages. Such Latin geminated consonants were preserved in the Romance languages developed in such geographical areas, whereas in other Romance areas they became plain consonants. In particular, they were preserved in Italian, which comes from 14th century Florentine.
The languages that have preserved original Latin geminated consonants have been found to be quite prolific in creating new ones by assimilation. Assimilation is a phonetic phenomenon in which a vocalic or consonantic segment changes to resemble an adjacent segment. For instance,
Latin SEPTE(M) > Italian sette, Sardinian ['sεt:e], Sicilian ['sεt:i].
Since you are interested in comparing Italian to other Romance languages, Catalan has also
produced geminated consonants by assimilation (this occurs more frequently in some Catalan varieties than in other ones). Even if Catalan shows degemination in inherited words, some geminated consonants are present in Latin borrowed words. For instance, [l:], written <l·l>, is present in col·legi and cel·la, Catalan words borrowed respectively from Latin COLLEGIUM and CĔLLA; [m:] is present in immanent, borrowed from late Latin IMMANENTIS; [n:] is present in innovar, borrowed from Latin INNOVARE; etc.
An interesting approach to explain degemination in stops, which is also valid for understanding the reason of consonantic lenition, is given by Alkire and Rosen in the book cited below. It has to do with the articularoty effort made when pronouncing consonants. For instance, if you are pronouncing a stop or a fricative after a vowel, the articularoty effort would be stronger in the first case because you will have to completely stop the air flow, whereas when articulating the fricative you will allow air to pass through. When pronouncing geminates, such articularoty effort would be strongest because you not only will have to interrupt the air flow, but you will need to prolong the interruption. A tendency to relax such articularoty effort is in some sense "natural".
But the mechanism of assimilation is really so frequent in languages that I would say it's also in some sense "natural" that new geminates are produced when pronouncing certain two consonant clusters. It seems to me that Italian and Catalan have in a certain sense evolved in opposite directions in relation to Latin geminated consonants. Italian has preserved Latin geminated consonants and then has produced new ones by assimilation. Catalan instead has lost Latin geminated consonants, but has acquired new ones by assimilation and then has reintroduced some etymological geminated consonants in Latin borrowed words.
References:
- Pietro G. Beltrami, La filologia romanza. Bologna, Il Mulino, 2017.
- Lorenzo Renzi and Alvise Andreose, Manuale di linguistica e filologia romanza. Bologna, Il Mulino, 2015.
- Ti Alkire and Carol Rosen, Romance Languages: A Historical Introduction. Cambridge, Cambridge University Press, 2010.
- The Oxford Guide to Romance Languages, ed. by Adam Ledgeway and Martin Maiden. Oxford, Oxford University Press, 2016.
- Daniel Recasens i Vives, Fonètica històrica del català. Barcelona, Institut d'Estudis Catalans, 2017.