Wednesday, May 31, 2017

What's In A Picture?

Last Wednesday morning someone emailed me the below picture from uBeam, asking if I was going to be writing about it. A few seconds later, I realized that I probably wasn't, as really there's not much new to say - anyone who understands hardware and business knows what's going on, the rest don't want to be educated or are investors. I also realized a couple of days later that pretty much no-one else cares either - I only had a single journalist call me about it and after giving him my opinion, he basically said that he was just keeping his notes up-to-date and editors weren't at all interested in uBeam. Since then, I've had a few questions from the tech side, and I noticed that the EEV Blog is commenting on it, so given I've a couple of spare hours tonight, here's my take. (For those who have asked, I'm in the middle of a piece about the recent Silicon Valley: A Reality Check blog post that got some attention a couple of weeks ago and will get it up soon, honest - read the original if you haven't already) 


I'm going to split this into three sections - my reaction, the tech, and the business implications so people can skip bits that aren't of interest.

The Reaction
What we have here, I'm going to assume, is a uBeam transmitter (large box with hexagonal tiles up top) and attached to the phone a receiver (the black brick in the bottom right). What was my first thought in seeing this? (After finishing laughing that is) Clearly they aren't any better at handling publicity than they've been over the last couple of years, with the PR firm clearly so asleep at the wheel they don't even know they're being ridiculed. Given they can actually manage to turn a positive article from a journalist into another piece pointing out the ridiculousness of typical day to day life in the company, it's no surprise that this quality a job was done here.

This, I believe, is the first official public unveiling of a setup that's been promised almost every year since 2011 or so, from a company that's raised $25 million. Is there simply no realization of how bad this looks, or is there an absolute lack of shame or care? Your big reveal should amaze and wow, it should scream that you are delivering, at the cutting edge, that you know how to get the details right from the finest detail to the grandest strategy. What this says is "Hey Granny, do you think I'll win the school fair with this?" not "Thanks investors for the $25 million, we're about to change the world!"

How unprofessional does this look? First thing is that they don't seem to be interested in tidying up the lab and staging before taking a picture to send out publicly, with bags of paper, white board, old carpet, and generally an unimpressive setup visible. It's like leaving your dirty underwear on the floor when your partner's parents come round for the first time. It makes me wonder if, like President Trump impulsively sending out Tweets in the early hours while no-one is minding the store, uBeam staff came to work in the morning to see the FB feed and said "we posted what?". If any of that team have been laughing about 'covfefe' then perhaps it's out of sympathy for the Whitehouse staff and what they have to go through every day. I read the goings on with President Trump and his staff and every day I see yet another corollary to working at uBeam - just in that case it's actually about something important.

Allow me to give uBeam some suggestions as to what to do next time. Begin with "decide the image you want to present to the world". Is it "sleek consumer design that Jony Ive would be proud of" or "cutting edge sci-fi level tech", for example, then setup the situation to reflect that aesthetic. For the former, put it in a sleek case that you've built with impeccable industrial design, you know, kinda like the one that was shown at the Upfront Summit last year? Any reason you couldn't use that? Put it in a user setting with happy people pretending to charge their phones, at the office or at the coffee shop. For the sci-fi side, get it in the lab with oscilloscopes and other equipment arranged in an impractical manner that would never actually be used that way but looks really cool to the average Joe. Hey you could even take that last setup to the next step and even show the voltage and current at transmitter and receiver!

Whatever you do, decide on a marketing theme that can sell and take the time to do it justice.

Along with the pics, actually put out a press release that says something. Take a look at Energous, they're perpetually 18 months from product after a few years and yet they put out professional press releases. You've a model to work from - take a look!

And one final suggestion - and it's not like anyone would have ever told you this before - when you're doing wireless power, don't show any wires in pictures of things being powered!

Seriously - I've not worked there since 2015 and I'm embarrassed by this.

The Tech
Now some time was taken to do a little blurring on the pic - you can see over the box that photoshop had a blurring filter applied in a circle over it. Perhaps this is to obscure that it's likely off-the-shelf components from Murata as I noted in a previous blog entry. (It could be an image overlaid to truly obscure and mislead, but I don't think so).  That's a bit strange, as I have pointed out in other posts that there's no issue in putting a mesh over the front to obscure what's behind, those Murata devices have it done to them by default. That mesh would also obscure the screws in the plastic and make it look at least a little more professional.

Blurring aside, the transmitter is now in a hexagonal arrangement, unlike the regular grid seen at the Upfront Summit, with 7 hex panels each around 15 elements across, so 45 or so elements top to bottom. If it is the Murata 40S4S then that's 45 cm top to bottom, and around 1600 or so elements total. At $3 each, there's $5000 of transmitter parts right there, and those are parts that sell in enormous volumes to car manufacturers so there's not a lot of room to lower the prices further. That implies a $15,000 transmitter, minimum (typical 3x markup from COGS to sale price). The box is maybe 3 to 4" deep, and it's positioned so you can't tell if there's a ton of electronics or power supplies sitting behind it - there were a lot of electronics in a large box in the Upfront demo, perhaps they've been downsized and rotated to fit in the box. Regardless, it's a pretty ugly box and nothing like the prototype that we were given a 'sneak peek' of at the 2016 Upfront Summit - what could have happened?

The hex pattern is interesting, and I wonder if it's been setup to beamform along the center line only only, no steering, in an annular array manner (concentric rings). This would really simplify the need for electronics, but if there are bad grating lobes (as you will get in an array where the pitch is larger than the wavelength), you can probably charge off to the side anyway as there's uncontrolled energy going in lots of directions. Basically, it doesn't seem the pitch is improved to allow for better steering and any real control - a big issue for safety in my opinion (beaming energy, you kinda want to control that). It also would not be representative of an array capable of steering in arbitrary directions.

The phone is attached to an enormous receive case that could be described as a 'brick' - it looks to be about 1.5cm thick which would allow for a number of the Murata transducers at about 1cm thick, along with an electronics board. Now an advantage of such a large box is that it will shield the MEMS gyros in the phone from vibrations which can damage them, but I doubt that's what it's there for - it's simply the smallest that can be made with Murata commercial transducers.

It may be that you've heard someone say that'll get better with "Moore's Law for Transducers", implying the transducers get half the size or twice the performance every 18 months. On the face of it is a bit silly as Moore's Law refers to the density of transistors on silicon and has nothing to do with ultrasound, but when you think about it and dig deep it's even sillier when you realise that the performance of ultrasound devices is generally tied through the laws of physics to particular device dimensions. Given those Murata devices were released at least 2 to 3 years ago, shouldn't there have been some major improvements to them by now?

But the phone is charging! Errr, well it shows 100% charge, but not that it's charging, and as before no idea of voltage, current etc that we really need to know the charge rate, nor of the overall efficiency of the system. Also, from memory, at >80% charge level the iPhone still shows charging even when less than 250mW are received.

So what do we learn from this picture? Not much other than there's a rearrangement of the previous demonstration, it's still apparently off the shelf parts, the receiver case is enormous, and there's no-one experienced in charge of publicity at the company to put out good pictures. As before, there can still be power received, even in the low 10's to 100's of mW, but as engineers and physicists have been saying all along, it's not transmitting power via ultrasound that's in question, it's can you do it at a useful amount, safely, in a practical way, at an acceptable efficiency, with hardware at a reasonable price. This still answers none of those.

The Business Side
As noted above, this looks like demo hardware, not even prototype, and still hasn't been shown working or efficiencies given. Can it work in a practical situation like an office or a coffee shop, or under standard use cases? What's the efficiency? Can it steer? How does it know where the phone is and track it? Most importantly, is it proven safe? Is it even legal at the dB level in most countries?

Reaction on Twitter seems pretty muted - in a week there's been a whole 3 replies and ~100 'likes' which is pretty telling. Seems the journalists know their audience...

What's really interesting though is that uBeam have not yet announced a new funding round. It's near 4 months since the Upfront demonstration, enough time to have completed a funding round with such slam-dunk technology. I'm saying 'no funding (yet)' as there are no new job ads, no publicity, and if one of the 'big guys' who a company will already have had come through to price the round were interested, they would not allow pictures of the tech to get out, especially if it's an Apple or a Google. Four or five months into fundraising things are starting to get stale, everyone knows that the first guys you spoke to haven't come up with terms (or acceptable terms), and that holding out will just get the company more eager to deal. Even with reducing burn rate by shedding senior staff and closing offices, runway only buys you so much in this type of situation. It's getting close to summer as well, and VC's are notorious for disappearing for July and August.

I have been expecting it announced soon, since there's no shortage of dumb money to go around these days, as when a 'low toxin butter-coffee' company can raise over $19 million, it seems anything will get funded. Perhaps they're holding out for the best valuation and getting that $100m round on an Energous-beating $400m valuation?

Who knows? And, from the public reaction to this picture, who cares?


Sunday, May 14, 2017

Recent Investments in Consumer Ultrasound

Earlier this month a consumer ultrasound company received a large series B funding round, taking in a further $23m to bring their haul to around $40m in total. It wasn't uBeam, it was UltraHaptics in the UK, who are using ultrasound to induce sensations and feelings at a distance (Haptics means 'relating to the sense of touch'). Think touching a 'key' on your phone when the screen vibrates or there's a buzz, even though it's not feeling like a key, the presence of some form of feedback tricks your brain and lets you know something has happened.  

It's likely a good complement to AR/VR and while it won't ever feel like solid objects or real things, the fact that there can be some form of feedback is a huge benefit. Imagine feeling something on your fingers as you touch a virtual keyboard, or sensations on other parts of your body as signals to interact with the virtual world around you. 

From the UltraHaptics pictures, it looks like they are using Murata MA40S4S car parking sensors (at least that's what they show), just many of them together in a square array (it's what's under his hands, just in front of the laptop). These are commercial off-the-shelf parts, and not ideal for a phased array, but unlike uBeam who also appear to be using them, are unlikely to be working at a power level where this becomes a safety issue. An array like this can be controlled to send beams in directions controlled digitally, but also as a receiver to allow imaging of the surroundings.


UltraHaptics seems to be working with the automotive industry according to the TechCrunch article, which surprised me, as there are fewer options within the car for that kind of feedback. What there is more demand for is sensing, both of passenger location within the car, and sensing close in around the car at low speed such as during parking. Other sensing systems such as LIDAR aren't always best close in, at a few meters or less, and ultrasound can do that job - essentially it's a high-fidelity version of the reverse sensors many cars have these days. I do wonder if they've found that there's another more lucrative application for their technology.

Another company in the ultrasound haptics space is Emerge, based in LA, and they have to be pleased that large investments are going into this space. Interestingly, Emerge has been on uBeam's radar, with rumours of a 'Cease and Desist' being sent their way for having the audacity to hire an engineer previously employed at uBeam (one wonders if the other 19 or so companies now employing the entire first group of uBeam engineers will also receive such letters). It's hilarious that anyone would be naive enough to think in California that a non-compete or restrictive practice could be placed on the employment of any engineer, and might indicate some desperation on their part. With Energous covering the IP space in wireless power (RF and ultrasound) as well as multiple applications such as communications, apparently few uBeam patents in the pipeline, and Emerge and UltraHaptics cleaning up in the haptics and possibly imaging spaces, that there is anywhere for uBeam to pivot to if (when?) the wireless power market proves unattainable.

Overall, I'm glad to see investment in this area - there are challenges for these companies but definitely some interesting opportunities - but also that there are engineering companies out there just quietly getting on with the job of building technology and delivering products.

Saturday, May 6, 2017

Raising Capital for a Startup: Convertible Debt

You have an idea, it's awesome and will change the world, something like a juicer or a toaster oven. Right now only you can see the potential, and you just need some funding to get started. You can use savings, but ultimately anything that's going to be amazing needs cash either to scale, or to get to market quicker - how do you raise that money? 

You can go to a bank for a loan, but they'll ask about sales, revenue, and profit, and seeing as you have none, they won't talk to you. So instead you go to an angel  (a rich individual or small group of rich individuals) or VC company, and offer them equity (shares) in your company in return for the money - but you still don't have anything, so how can they value your company to determine how much stock they should own? It's a large and expensive exercise to work that out (and you have no money to pay that with), at the end of which you may not think it's fair or viable for your needs.  

So with no revenue, product, idea of actual value, how can you reasonably raise money? A common way this is done is Convertible Debt. Convertible Debt is a hybrid between a loan (debt) and equity (shares) that tries to keep things simple in terms of the paperwork to get going, and puts off the tricky bit about valuation until later when there's more information to base that on.

Here's how Convertible Debt works - the company and investor agree on an investment amount, say $100,000, which the company will use to further the business/product. This is usually after a few rounds of meetings and presentations, where the founder has shown a basic pitch deck, presented a plan and a vision, and been vetted to some degree by the investor. Each Convertible Debt note can be different, I'm presenting a common version, but expect every one to have its own idiosyncrasies.

This money is given as a loan, same as with a bank loan, complete with an interest rate and repayment period - for example it might be at 3% interest, with a 2 year repayment timeframe. At the end of those two years the debt, along with interest (which has been accumulating all that time), is to be repaid in full. There are no debt payments made during that time, unlike a regular bank loan. 

That's pretty straightforward for a loan, but there's the equity part - if there is a "funding event" at some point in that 2 years (usually defined as a certain total amount of money raised) then the loan converts instead to equity (shares). Basically it starts as debt, then converts to equity, hence convertible debt.

How does the 'convertible' part happen? At some point a valuation of the company can be performed - but here the company has (say) 2 years of work behind it, perhaps a product prototype or early customers, and the VC firm putting in the Series A money (sometimes called "institutional money") has the capability and reputation to place that value on the company.

So, let's say that $100,000 of Convertible Debt was put into a company that a VC later values at $10,000,000 pre-money, puts in $2,000,000 for a $12,000,000 post valuation - that converts based on the pre-money valuation (usually, but not always) so those original investors get around 1% ($100k/$10m, ignoring interest) of the $12m company. They gained about 20% ($120k) on a pretty risky bet over 2 years, so it's a decent return but not earth shattering considering they basically made the company possible, were likely to lose everything, and the VC now has ~17%. 

This is a little unfair to the original investor, so this is why there is also a "discount" or "kicker" in the Convertible Debt agreement, where there is a discount on the price, often around 20%, so they get a bit more. With a 20% discount, they'd be getting 1.25% of the company ($100k/$8m) - now they've made a 50% return in value which is better, but still not that great, especially when that money isn't liquid and they still have the risk of future rounds of funding and it all going wrong. If the company happens to go stellar with that initial money, say a $100m valuation, then the investor gets an even smaller % of the company - that's a great deal for the founders and VCs!

Sophisticated investors in Convertible Debt often ask for a cap on the note (since it's debt, the term 'note' is often used for Convertible Debt). In the case above, they might have a note with a cap of $2m - in that case if the valuation goes over $2m at the Series A, the conversion of debt to equity is calculated at the cap - so it's ~5% ($100k/$2m) and now they've got a (paper) 500% gain, which will make them much happier. (Whether cap and discount both apply is down to the details of the note, sometimes it's just one of them).

So Convertible Debt has the benefits of keeping things simple in the legal papers (it can be done in a couple of pages), puts off the tricky aspect of valuing a company, and allows for the upside of equity in a growing company if things should take off. This is why Convertible Debt is a common financing vehicle early in a company, often in what's called the 'Seed Round'. 

What are typical terms? Usually these notes are for amounts in the $10k or $100k ranges (by definition it's a small company trying to prove things out, with a 'non-institutional investor'), but sometimes go into the millions. Interest rates are usually nominal, say 1 to 5%. The discount is also variable, but 15 to 30% is not uncommon. The term is a bit trickier, how long to make that? Well, you need to be actually able to do something with the money, and then with your new prototype or product go to a VC and the process of raising a Series A. Conventional wisdom tells you that if all goes great, a Series A raise takes 3 months, and it generally doesn't go well so assume it's 6 months. Basically, however long it's going to take you to get to something worthwhile, plus 6 months, is how long you want. For example, if you think it's going to be 9 months to a year to get to prototype, don't make your note term shorter than 18 months, and you probably want to give a little headroom in there for things going wrong. I've rarely seen a single year as a term, but I'm more a hardware person and those projects take longer, with the bulk in the 18 month to 2 year range - longer than that is rare as running a company for 3 years on convertible can be tough. Terms can also be conditional - that is they change with certain events - for example the discount may increase at certain points during the term of the note, starting say at 15% on a two year note and increasing by 2.5% each year. A founder may offer these terms to entice the investor to give a longer term, or the investor may want to encourage alacrity on the part of the company.

Most times both investor and company assume that it will be a clear situation of successful funding and conversion, or that the company has gone under, but if it's in limbo or limping along as often happens, things can be uncertain. What happens if things go wrong?

The most common way is for the term to expire and there to have been no funding event - the investor is due the loan back with interest, but no institutional investor agrees the company is worth funding, and we assume there is no money in the company to repay it. What happens then is the same as with any debt - debtor must come to an agreement with lender as to next steps, and this could be anything from bankruptcy to a renegotiation of the debt. With smaller amounts, both sides might just ignore it and pretend it didn't happen, with legal costs likely to outweight the investment amount, but the larger the amount the more an agreement needs to happen. Bankruptcy doesn't make sense, usually, as driving the company under ensures no chance of future success, and they are unlikely to have significant assets to liquidate to pay the investor back. It makes more sense for the two parties to come to an agreement, for example extending the term for another year while increasing the discount by 10%, and so it's usually recommended that if money isn't already in, start talking to your Convertible Debt note holders at least 3 months prior to term about what happens next.

Regardless, any institutional investor coming in for a Series A will want that paperwork cleared up before their money goes in, as they don't want a lawsuit or trouble later on.  This leads to some interesting situations where the Convertible Debt investor can start to demand beneficial terms from the company and hold up the Series A, or the company can demand the note holder give concessions like the discount rate or they won't go ahead (it's a "give me what I want or I shoot my company" tactic but I've seen it work). In those cases, it's down to bad blood between Convertible Debt investor and founder, and if it wasn't before it certainly is after.

One interesting permutation I've never seen play out is if a company does a convertible note between institutional rounds - for example between Series A and B when there was an original valuation and equity investors. It's odd, but it does happen, as LA Business Journal's Garrett Reim notes this is a route uBeam opted to follow when they took an (up to) $15m Convertible Debt round in July 2015 after a ~$10m Series A in summer 2014. This leads to a lot of possible weirdness that may or may not occur depending on how uBeam's fundraising for their Series B (which they must be deep into, now nearly 2 years from last fundraise) plays out.

First weirdness is that the Series A VC investors, who all get Preferred Stock that guarantees them paid out first from any money, are very unusually in the queue behind the convertible debt for being repaid, as debt always takes priority in any liquidation. This puts the lead investor in the position of desperately wanting a conversion from debt to Common Stock so they can take priority again.

Next is that the convertible round size was likely based on the then valuation of the company, with the expectation it would rise in the time between loan origination and maturity. If the situation arises where the Series B is a down round (lower valuation) than the Series A, or even similar, then the Convertible Debt investors will end up taking a huge % of the company even before the shares of the new Series B investor dilute the company further.

Lastly is that if the company had a valuation prior to the Convertible Debt investment, then there is the assumption that there is actually some value there, be it product, customers, IP like patents, or even remaining cash in the bank from the original Series A. A Convertible Debt investor at the note maturity may decide that liquidation is the best outcome for them, especially if they can get paid off first (perhaps they know the company has more cash in the bank than they are owed), and they do not think the company has much of a future. This can lead to an acrimonious situation as the investor plays 'hardball' with the company - that's not going to happen at the $100k level, but go past $1m and things are different. An element of that played out with Theranos when one of their investors, PFM, sued for their $96m investment back claiming fraud, when they know Theranos had $200m in the bank - basically taking the money before lawsuits and time removed the potential for any return.

Overall, Convertible Debt is a well understood way of raising money in the very early stages of a company, with simple terms and paperwork, but it can lead to some very difficult situations should it not convert at the end of the term.  There are other options and variations, such as SAFE, but in the interests of simplicity, I'm focusing just on that.