There is no one-size fits all strategy. However, there are factors that will significantly impact the onshore/offshore manufacturing decision.
The most important factor is the relationship between the production volume and the value of the product. Are you creating a low value, high volume product (such as a consumer electronics product)? Is your total cost of goods under $50? Are you ready to commit to making at least 10,000 of your product in the first year of production? If so, then offshore manufacturing is probably the right path. On the other hand, if you are creating a high value, low volume product (such as a medical device) with a healthy cost of goods sold, or COGS (e.g. $1000) and a relatively low volume (e.g. 5000 a year for the first year of production) then you should seriously look into manufacturing in the USA.
A second factor is the level of intellectual property protection you are looking for. Are you working on a government contract? If so, you are definitely manufacturing in the US. You may have to choose a CM who is a certified defense contractor, where everyone on staff is a US citizen with security clearance. That is an extreme case, but if you have IP you are looking to protect, you may want to rethink offshoring. Case in point: an exact replica of one of my prior products popped up in Asia a few years after we started producing them in Southern China. Somebody somewhere leaked the technical package. There wasn’t anything anyone can do about this, since IP laws are not exactly enforceable in China. If this is a significant concern, manufacturing in the US will give you much better peace of mind.
A third factor is the precision and level of invention needed in process development. Does your product have lots of moving parts? Do you need high precision parts? If the answer is yes, then process development is a lot more significant. The counterexample is commodity products like laptops and mobile phones. These involve fully validated and low risk manufacturing processes. Developing and honing a complex production process is far more easily done if the CM is close to the R&D center.
As a general rule of thumb, military products (usually made in low volumes) need to be made in the US. High value products such as medical devices and luxury goods (made at any volume) are comfortably producible in the US. Low volume, low value products are suffering. While making these products in the US makes the most sense, it is hard to get the COGS to an acceptable level. This is because much of the cost is in the setup of the manufacturing line. This is often a fixed cost regardless of quantity of products made. At low volumes, this setup cost gets passed on to the startup in the form of a higher COGS, significantly impacting their gross margin.
Not having existence proof of a viable COGS makes it very hard to derisk a device sales based business model, which only works at scale. This is a very interesting problem, with several groups working on solutions in the US and in Asia. It will be interesting to revisit this issue in a couple of years to see if things have improved.