"Having trust" and "operating as if you have trust" are two different things. I can trust my management with spending authority, but I still have periodic audits.

Trust is not binary.

Blockchains, via crypto, remove uncertainty from medium-trust relationships that would otherwise incur overhead due to counterparty risk, allowing one to replace it with the risk that e.g. SHA256 or ECDSA signatures have been broken, which is usually orders of magnitude lower than squishy human being behavior risks. (Presuming participating organizations have a way of keeping their keys safe, which is expensive but solved.)

That overhead is significant, and entire multibillion dollar industries exist around validating those trust relationships (accounting, auditing, settlement, et c).

Imagine if you could replace them with software that doesn't have to take anyone's word for it.

That's a big efficiency win.

Could you give a concrete example of how blockchain technology can reduce costs associated with accounting, auditing and settlement? Just a single, concrete example is all I'm looking for.

Securities exchanges usually have settlement as a distinct step from trading. Blockchains as title make settlement faster, easier, and less expensive.

With distributed blockchain enterprises can make use of each others assets without actually a single party dominating, and with zls layer over distributed block chain they can make use of each others confidential assets without actually disclosing the secrets. Like a customer of one bank can prove his credit history to another bank without actually giving his credit history. A patient of one hospital proving his medical history without actually disclosing medical records. Supply chain efficiencies without disclosing trade secrets. And many other efficiencies and asset sharing.

The problem with these examples is that the literature is filled with alternative protocols that accomplish the same task without the deadweight of the blockchain. This seems to be a solution in search of a problem with the added cost of "let's make it anonymous because we are cool" without tangible upside. If you start from the assumption that we are all mutually distrustful, adversarial entities it may make sense, but this does not accurately describe any of the enterprise relationships I have ever been involved with.

To some extent we are mutually distrustful, adversarial entities and to overcome that we employ contracts, compliance , patents etc to minimize the distrust and this adds overhead of lawyers, lenghty paperwork etc. With blockchain we can solve the trust problem via crytographic means more efficiently.

It seems to me that the problem with this answer is that we have already paid the up-front cost of these offline systems over generations and so you are supposing that the switching cost is less than the ongoing maintenance of what already exists. There may be efficiency gains to be realised somewhere, but unless there are incremental and quantifiable advantages at each intermediate step from here to there I can't see this existing outside very small (albeit potentially lucrative) niches.

Blockchains are ponzis. You need them so that you can offload yourr risk onto the community and exit your project profitably. Non-blockchain investments don't offer that feature.

Yes they do? Even fiat cash has that feature. So do stocks...

To clarify: I'm not arguing that blockchain technology is useless for enterprises. I'm requesting concrete examples of blockchain technology solving a specific B2B-issue better than existing technology. All I've heard so far are vague, theoretical examples, while no one seem able to point to a concrete B2B-issue, and explain how and why blockchain technology solves this issue more efficiently.

Enterprises are using blockchains based on trust rather than proof of work. The trust is often manifested as some level of consensus among nodes and the tendency of a node to offer a 'dissenting' value is a basis for establishing the degree of trust. For example 70% of trusted nodes concurring. In some protocols each institution can maintain its own list of trusted nodes. In other protocols the list of trusted nodes is agreed among institutions. Like all engineering, it depends.

They use blockchains because it is a fast way of conducting transactions relative to existing infrastructure that is architected based on incremental automation of manual processes. Transactions on a distributed ledger can clear quickly relative to standard send->ack->transmit->verify protocols with human scale execution times.

Is it all rosey and perfect? No. There are bugs in the system. There are bugs in every system including the existing system.