A lot of it depends on the hardware you're going to use. Assembling a bunch of off-the-shelf components into a new idea (ref: Chumby) is going to be much cheaper than developing your own custom asic's (ref: SiCortex).

My experience over the course of a couple of hardware startups doing more off-the-shelf approach is it would take you about $20-$35MM in total funding for a typical hardware startup doing something semi new/radical.

The hardware design itself can actually be relatively cheap. Packaging (custom tooling for cases, etc.) can easily cost you $20K-$300K (ie: as much as a big Angel round for some software startups). Manufacturing costs are usually a function of volume commitments and timing. So, you are going to be incented to commit to larger manufacturing runs than you are comfortable with to keep the per-unit cost low. Inevitably, after you do this you find there is some trivial but important item you missed, so you end up with re-work costs, or if you're very unlucky a lot of scrapped inventory.

To answer your Tablet PC question, my guess is that you're looking at $10M. There is not a lot of invention there. You'd pull together a lot of off-the-shelf components into a new form factor. Most of your early costs would be sucked up by the design and tooling of the casing and overall industrial design. I'd venture that you'd be $500K into it just for ID and an ME to layout the (presumably) injection molding designs for the case and first mold. Budget another $150K ish for UL and FCC certs (your outsourced factory can help you with a lot of this process). Then you're likely to go through probably $250K in the initial hardware designs and single-unit prototypes. You'll generally need or want 1 prototype for each developer (probably 5-8 in this case), plus another 6-12 for QA, and another dozen or so for demos and for employees to carry around and just "use" out in the wild. If you're making a tablet with an $800 price point, your production BOM is going to be around $200, and total manufacturing costs around $350ish, but your first prototypes will cost you $1000ish/ea at first because a lot of it will be work done by hand, etc.

So, thinking purely off the top of my head, you'd probably take a $1M angel round, a $4M A round and a $5M B round. Since this is a semi-proven CE device and not something radically new, you just need to build it and get it out there, not convince the world why they need it.

Wow. I was part of a small team that put an industrial product together, and we didn't need nearly that much money. It was not a startup, so the cash flow from existing business easily supported the one-year timeframe. (We did the design in our 'spare time' when we weren't busy with day-to-day work.)

No industrial design (very utilitarian product, simple molds), and we had someone with a bit of molding experience. Had to go outside for mold design and construction, of course. We went with hard steel, single cavity, half a dozen molds all in, but I'd certainly go with aluminum (for the initial production) if I were to do it over again. I'm thinking we spent just over $100k on mold stuff before we had our first test shots. Talk about an exciting UPS package! ;-)

Our product didn't require much certification, so we saved there as well.

Not only are the first 100 electronic widgets much more expensive, it (was?) hard to get them made because our usual suppliers weren't set up for a low-volume job, and we didn't have existing relationships with the more proto-oriented CM. I did a lot of soldering that year.

It would be much easier today, especially with so many cheap dev/eval/reference kits on the electronics side and the rise of quick turn mold shops such as Quick Parts and Protomold.

Hardware is a real blast. Nothing quite like seeing those molded parts once they start coming in in pallet quantities!

I've put small/niche little projects out the door with less than $500K of my own money, and I've seen hardware startups blow $40M with nothing to show for it.

I didn't give the full breakdown, but the numbers I used above also take into account that you're building an actual company. Where you're paying people vs. having people work in their spare time, and you have other overhead (rent, etc.). Also accounting for a moderate PR and tradeshow budget, which would all be typical if you're trying to build a Consumer Electronics product, even with the ref. designs and such out there today.

There are a lot of variables involved, and a lot of places to save money. Prior experience on your team also works in your favor.

Wow, that's a lot. Is that how much you reckon Anybot is spending? Would it be a lot cheaper to simply sell your hardware as a kit (less certifications, etc)?

The cost savings from requiring user assembly are much smaller than the market loss from limiting your market to folks who are willing to assemble.

Not sure about Anybots. Only remotely familiar with them through some mentions here. A quick (very quick) bit of research looks like they're part startup and part experiment in terms of investing in the robotics space overall.

Most of what they're doing is work, not invention, so they should be on the cheaper end of the spectrum. They still likely have a relatively high BOM and a ton of tooling costs for custom parts and such. Additionally, they're going to have to spend a lot of money on PR, tradeshows, and demo equipment to get it anywhere near viable (assuming it is meant to be more startup than experiment).

It's also too soon to tell, a small bit of fate could swing something like that wildly in either direction.

If anybots were a true stand-alone startup (and for all I know, they very well could be), I'd put that at around a $20M net (aAB rounds) investment to get chugging. Could highly likely need a C round to scale out, depending on distribution model (do you buy them online, at Best Buy, or both?).

Great info, nice job.

How would you recommend sourcing components?

Which components, exactly? :)

I buy most one-off parts-n-pieces through DigiKey.

I've done a ton of prototype PCB's with ExpressPCB. (expresspcb.com)

For CNC work, vacuum-form molds, and injection mold ME's, I keep some local folks on speed-dial. I often reduce costs here by doing the roughing-in of mechanicals myself. I have a moderately well enough equipped shop that I can do prototypes for cases, bezels, etc. myself. I will most often sketch something out freehand (I'm a HORRIBLE artist though), then make a wood model to see what it looks like "in person". Then (generally) do a fiberglass model (because it's cheap and easy to work with) of the part, check rough fitment, THEN go to an ME or CNC person to discuss costs and production times. I've done that cycle enough times now that I generally know how to make something "manufacturable". IE: little things, like complex curves for an injection molded part, can add a comma or two to the mold costs :) In any case, having a physical part to base your discussions off of makes things easier. Most of the stuff I've done hasn't warranted a bona-fide Industrial Design cycle (or if it was warranted, it wasn't budgeted :) )

The major costs would be the assorted tooling charges: - plastic moulds: for a tablet size probably $50-70k. I would recommend you get an experienced industrial designer to guide you through this process. Make sure if you get quotes from China that you don't give the 3D model for the quote - give 2D technical drawings with enough information for a quote but not for full production (your industrial designer should know this). - PCB boards: nowadays tooling charges are quite low. We use OurPCB.com for a smallest circuit board and the tooling charge is in the low hundreds. We source the important components and get them to ship it to the PCB makers. they then send the completed PCB's to us. - packaging: allocatae a few thousand to set up your box or plastic packaging.

Stage your production: - pilot run (say 10 PCB boards to get all of your component logistics sorted out). - 1st run (100 PCB's with 100 plastic shells and packaging) - production run (1000 PCBs, 1000 plastic shells and packaging)

Assemble, program and test your devices close to you (ie. your garage :). This allows you a) to keep your IP close to you and b) iron out any production bugs, processes etc. so you are ready to outsource the whole production and assembly (if you so wish).

Hardware as a rule isn't necessarily expensive. If you had a specialised peripheral you might be able to get going on a million -- maybe even a lot less if you can push functionality to software (eg a simple setup where the fpga or microcontroller does all the heavy lifting).

The problem is when you want to compete in a market with volume and wide distribution. Eg something like a tablet pc. Unless most of the idea comes from OEM parts then you need big money.