Shiro is one of the very few companies doing extremely accurate, precise and delicate CNC work in the knife field. Both Shiro and CPK exceed what CRK has done in their respective fields.
You wouldn't know it looking at Nathan's recent offerings, the fixed blades recently being relative simple compared to a folder but he is capable of more than what you see machine wise in the relatively basic current models.
I've had oh maybe 12 Shiros go through my hands this year, and disassembled and tuned probably 5 different Shiro models.
I've had over 30 folders from CRK. Tuned and disassembled at least 12 individual examples between the small 21, large 21, sebenza 25 and inkosi models. I EDC'd a CRK for years and did regular maintenance on them.
Without a word of exaggeration the worst of the newer Shiros is many times better than the best CRK I've had, a PJ 21 with an old raindrop Damascus blade from arguably the best CRK years 2012-2014.
If I put a handful of newer Shiros on your table and a handful of newer CRKs, the Shiro is to CRK as CRK is to Kershaw. The level of attention to detail, quality of finish, action of the pivot and the lock bar interface, execution of design, usability, steel selection are all massively superior in the Shiro.
Side by side; you can instantly see how the Sebenza has been designed to make it easier to manufacture. While the Shiro uses a far more complex design that results in a much more impressive, comfortable tool for my hand.
The only area I would say CRK is ahead of Shiro is the ease of maintenance. CRK provides a standard hex to disassemble thier knives, Shiro uses a proprietary fastener which requires a proprietary tool they charge several hundred dollars for to remove the fastener without damage.
While I do not know that Nathan can compete with Shiro at this point, I for one would love to see what he can create with his level of innovation and impressive interpretation of a design there are literally hundreds of similar versions of. His ability to translate a concept into reality is second to none.