Another glorious Palo Alto summer day and there I was, lounging against a filthy drainage pipe outside Jin Sho on Cal Ave. Face in my phone, I was eavesdropping on two guys in suits pulling information from a prototypical tech dude (beard, polo shirt), fishing for the real deal in Silicon Valley.
“mumble mumble biotech mumble…”
It’s the only word I can recall – I can still hear it, hovering in the hot air – because I was interrupted by a man I’d never seen before cautiously approaching, looking at me with the same shy, sideways and slightly bored do I know you smile we startup folk wear when meeting new people.
The first thing I noticed about Mike Henry, founding CEO of Mythic, was how little he resembled his picture on Linkedin.
“You have no idea how many attempts I had to take to get that stupid picture right,” he shared.
I knew what he meant – I’ve taken one good photo my entire adult life, snapped by a headshot pro in 2014, and have milked that thing for all it’s worth. I also knew Mike, having recently closed Mythic’s Series B – where Micron was an investor - was doing me a huge favor by meeting for lunch to talk about…stuff.
I’d been vague with my reason for the meeting. In fact, I didn’t have more than a concept of getting my new crew – Micron Technologies and our AI Fund – more involved with startups and drumming up ways to support Mythic.
Money is good, but oh – investors can and should do so much more!
The second thing I noticed about Mike was how relaxed he felt on the back patio of that sushi restaurant. Here’s a guy, a founder, who spent the last 7 years of his life exploring, trying, failing, building, trashing, recycling, pouncing, failing, building, building, building…and he’s alive. Mike and his team at Mythic are alive and better yet, succeeding, after years of relentless commitment to making it all somehow, despite the odds and startups headwinds, work.
I came into the meeting with no expectations. I left with the full expectation that Mike and Mythic are growing into something truly impactful and that their story, experiences, and lessons along the way have positioned them to take full advantage of their massive opportunity.
Here’s a couple excerpts from our talk on the patio:
Me (AB): OK! Mike – you guys originally founded as Isocline.
Mike (MH): Yes.
AB: So…why Mythic?
MH: I had ten minutes to come up with a name. I was writing government research proposals, so I grabbed a math book, flipped through the glossary, searched for available domains, and I thought “Isocline, that sounds kind of like Raytheon. Perfect, let’s do it.”
MH: Turns out the domain was not actually available. Couldn’t even get the domain.
AB: Well done.
MH: Yeah, so we had it for four years. Nobody could pronounce it.
AB: Isocline? It’s incredibly complex.
MH: And everybody asked, “what does it mean?” and I’d always give the same stupid story about how I had 10 minutes to come up with it. Very uninspiring.
MH: We hired a naming expert.
AB: Humph! Sorry, I almost just choked…on a grain of rice.
MH: It was such a fun process I almost want to use him to name my kids. We had all the founding engineers involved, we went through hundreds of different names and he made up a little story for each one on the spot, and everybody just resonated with Mythic.
AB: It was probably insanely valuable. I won’t speculate on how valuable.
MH: Great brands are so important. The only vestige we have of Isocline is t-shirts.
AB: I want one. Ok – next question. What was the original vision of the founding team? Were there any pivots along the way.
MH: Early on the founding team was interested in new modes of compute. There wasn’t a guiding market or application perspective. Just an overall sense Moore’s Law is coming to an end and what’s going to drive the chip industry for the next 40 years will need to be far more creative than what’s been driving it for the last 40 years. The earliest technology was GPS, our first contracts were GPS for the Air Force.
MH: And we came up with an analog compute method for that, which worked very well. Laura Fick, one of our founding engineers, also started porting the technology to AI and neural networks.
AB: This was in 2012?
MH: Exactly. The earliest investor pitch decks, I go back and read them sometimes. It’s so funny how bad they are, since I was trying to sell both technologies. We were literally a GPS and AI company.
MH: And their reaction was “what the hell are you talking about?” Low-cost GPS chips is commoditized - a couple players control the whole market. We saw AI as this total green pasture, billions of dollars, no leader, and we pivoted into AI and abandoned GPS. We literally made a GPS receiver that’s 50 times lower power and its just sitting on a shelf somewhere.
AB: I want that, too.
MH: Well, it needs a lot of work to get it going. But we demonstrated the technology. And it’s sitting on a shelf.
AB: Interesting. I have an equation for startup success - VCs get to think about stupid stuff like that. And its startup success equals founder squared luck raised to the timing. SS(t)=(F2L)t dt
MH: That’s a great way to put it.
AB: There’s a lot inside of this F quantity, founder quality, how much technology is inside of it. Luck – everyone needs a little bit of luck.
MH: Yeah, we stumbled upon the AI.
AB: Your timing though? Is breathtaking.
MH: We actually totally mistimed it, but we luckily had a good pipeline of government contracts to keep us afloat. If we’d raised venture in 2013, it would be way too early. The markets weren’t ready yet.
MH: Yeah, and we would have had these VCs on our board treating us as a stale, dead company. 2016 was perfect timing, and those government contracts kept us going as we continued to develop the technology and wait for the markets to evolve. We got the timing wrong but survived.
AB: Cool. Then your L quotient is massive.
MH: Yeah, our luck is higher. Literally, tiny little government contracts, these little bags of money we found in the most random places – we almost ran out of money so many times.
AB: What was the lowest you ever got?
MH: Two weeks.
MH: And that was six months before the series A.
MH: A last minute call to an angel investor saved the company. So, yeah – luck huge.
AB: Luck. Luck. Luck.
MH: Timing crappy, but we recovered.
AB: Your luck was just big.
AB: And your F value was big.
MH: But you gotta exploit the luck.
AB: That’s right.
MH: That’s a great equation though.
AB: Ha-ha! Gracias. Alright – next question. There is a lot of excitement around 5G… How do you see 5G impacting AI applications, particularly on the edge?
MH: I think it’s gonna be great for a lot of things, but I don’t think its going to replace compute right at the edge. If anything, 5G could increase quality of data going to local aggregation points. If you get a lot higher quality data going there, and you could need a lot of low-cost low-power AI compute there. If you need high reliability, 5G + AI is not the solution.
AB: Especially for mobile applications.
MH: Yeah, at those high frequencies it’s not as resilient as people make it out to be – it starts raining, and all of a sudden your compute goes away?
AB: Ok. Weird question. In your mind, what do you think will be the most interesting applications enabled by Mythic technology and products.
MH: I think it will be a while before AI spawns a whole new class of products, like you saw with the internet and mobile. It will happen though. In the meantime, you’re going to see existing products and features get significantly better. Cameras, robots, manufacturing, internet search, cyber security - on and on. Name any electronic product or enterprise system, there are 5- or 10-ways AI will make that product significantly better and more enjoyable.
AB: Got it. Last question, and apologies for the shameless plug but I’m curious to get your response. Micron – its an interesting company, and we’re excited for our investment in Mythic through our AI Fund. What excites you about partnering with Mythic on your startup journey?
MH: We’ve been talking to Micron for a while now, so we know some of the team. When we started mentioning Micron to our investors and partners we could tell there was a lot of excitement. For any product, the entire system needs to be 10x better, not just the compute. There’s so much expertise within Micron to help us with that goal– we see that as a big opportunity.
More fundamentally, AI processing – both at the edge and the datacenter - needs very capable memory and storage. I’m hoping for some very special products coming out that leverage the advantages both companies offer.