Dual Data Center Advice

I work for a university that has been doing networking since the 80s. As you can imagine, there are a lot of legacy policies and designs built into the network. A couple of decades ago they added a 2nd data center for "redundancy". This DC is much smaller than the primary DC, but the concept was the production box for a service would go in one and the devel box would go in the other. If the prod failed we could then bring up the devel box as the prod. Well, virtual machines has all but abolished that practice. However, we(as an organization) have deep rooted fears of moving everything to VMs. Our ERP database is metal and there is no plan on migrating it to a VM.

Another aspect is there is 0 routing in the Data Center. It's all done at the core. So we have VLANs spanned across both DCs(and a 3rd for backups). There is another issue in that every time I turn around management or clients are asking for new VLANs. We have over 255 VLANs on campus. We do LOTs of VLAN pruning. It's a nightmare However, campus design is a different topic.

My main goal is to remove as much spanning tree as possible and route into the data centers. Currently each DC is powered by a pair of Nexus 9372s with FEXs hanging off of them. This was a decision made a few years ago so that I could get rid of the 2960s powering the DC. I got what I could, when I could. We are now poised to start planning for the new budget year and I've been tasked with figuring out how to get 10G, 40G and 100G into the DC. Well, that means different switches. Management is wanting more 10G attached hosts.

My questions are as follows. What would you recommend in terms of hardware for the switches? Would you recommend moving to a spine leaf architecture and overlaying VXLAN? If so, how would I connect the 2 data centers? Can the VXLAN overlay work between 2 DCs?

I'm just very concerned that I'm barking up the wrong tree. Some of my management also has the mindset of "If it ain't broke, don't fix it." I'm also worried about implementation and how to transition between the 2. How much downtime will required. Any criticism is welcomed. Thanks everyone!