Do I have to acknowledge the company where I started a project when I later publish related research in college?

At least in my field, Acknowledgements are generally regarded a fairly informal mechanism. There are few rules dictating what must and mustn't go in there. However, given that mentions in the acknowledgements don't dilute formal credit, I see no reason not to be as generous, gratuitous and grateful as possible in an acknowledgements section. Its not like adding an author, it costs you nothing and makes you look like a better person. Why wouldn't you?

Also there is probably an argument that you spent time thinking and learning about the general problem domain while being supported by the company.

I recommend adding something like "The author acknowledges BigCorp PLC for first bringing the problem to the authors attention and [financial?] support for work on related approaches", to the acknowledgements.

What are the consequences if I don't acknowledge the company.

You just burn bridges if you do that (you basically behave socially wrongly). For future conferences, it will hurt your professional reputation

FWIW, any European funded H2020 project contractually requires such an acknowledgement (at least in some footnote). AFAIK, NSF or NASA or ESA funding -of semi-academic work- has similar rules.

But I feel that the only role of the company was to give me the problem statement.

In Europe, that role alone is worth many thousands (or even millions) of €. Reports like ITEA3 roadmaps (or future HorizonEurope call for proposals) are exactly defining problems and inspiring funding of research proposals. The people writing them are paid, sometimes full-time. I guess that Indian people writing Indian DST call for proposals are paid too. Very probably, if you are happy enough to get funded for your research by them, you'll be contractually obliged to acknowledge such funding institutions.

In addition, research motivated by industrial needs is, in this decade, more professionally popular than unfunded research.

Suggestion: acknowledge both the company and your professor (unless he is a co-author).

FWIW, I tend to believe that lack of institutional acknowledgement in past submission related to RefPerSys (a not-yet-funded research project in AI) have been negatively evaluated.

I'd like to add to the so far excellent answers:

• Increases the reputation of the company, because they contributed to your research - even if you think in this case it wasn't much.
• Increases the reputation of the field, because the paper shows there is interest for it in the private sector. This might not be important in this subject, but a lot of fields seem to have to practical applications. Showing at least one company was interested disproves that assumption.
• Increases your and your professor's reputation, and shows you are both available for practical research.

And - as already pointed out - this has no negative side, no cost, nothing.