California startup Luminous Computing just announced Michael Hochberg, PhD as the company’s incoming president. His new leadership role in the machine learning systems company comes on the heels of a $105 million Series A round of funding. Luminous seeks to overcome the shortcomings of today’s existing supercomputers by delivering scalable AI systems using silicon photonics to build more power-efficient and bandwidth-dense data links. I caught up with Dr. Hochberg just after the announcement to learn more about how he’ll approach his new leadership role.
You’ve founded four companies in the optics and photonics space, two of which were acquired by Cisco and Nokia, respectively. You’re now serving as president of Luminous. With such success in turning startups into viable companies, what are some of the guiding principles you follow that will help elevate the success of Luminous?
One of the things that I've learned is that every startup is unique. I often get questions from entrepreneurs about the 'right' way to build a company. There are some dynamics that play out again and again, where there really is a 'right' answer. But for most of the biggest, most impactful choices, it's really an aesthetic choice that the founders and the execs have to make about priorities, risks, and values. I think the most important thing that you get from experience is a sense for what things are common across a lot of different circumstances. Here are a few that I've seen play out again and again:
• Getting absolutely first-rate people is crucial. With excellent people, very difficult problems will solve themselves, without intervention from leadership. With mediocre people, even trivial problems become impossible to solve. Running a great interview and bringing in the very best people is do-or-die for any startup company.
• As a company grows, it becomes more and more important to explicitly define roles, responsibilities and deliverables. When you make the transition from a small organization where everyone knows one another very well (say, up to 20 or 30 people) to one where people don't personally know everyone at the company (50 and up), people need to have clarity on where to go and who is responsible for different areas of the project. The roles tend to become more specialized, and it becomes necessary to have people who explicitly specialize in being generalists, and seeing multiple different parts of the project. Delegating responsibility and authority together is really important.
• Having a clear vision for what you want to build and why it is essential. Spending a lot of time talking to customers and understanding their pain points is not optional. The founders here have done a really great job in this regard and have identified some unique insights into things we can do to really improve our customers' experiences.
• Having committed, long-term investors makes a world of difference. Again, the founding team here has done a wonderful job of selecting investors who are in this for the long haul and who share their vision for building revolutionary, moon-shot technology.
Having served in leadership positions at each of your startups — first Simulant, then Luxtera, Silicon Lightwave Services, and Elenion — how do you establish priorities when guiding a company toward success?
For me, one of the things that I find is always top of mind is bringing excellence onto the team. Hiring first-rate people makes it easier to hire more first-rate people. So I spend a fair amount of my time on recruiting. I personally talk to every candidate who we bring onto the team. My job is to bring in excellent people, and make sure that they have the right amount of structure and the right resources to generate brilliance.
Customer focus is also crucial for me and is one of the reasons I joined Luminous: CEO/Co-Founder Marcus Gomez literally used to be our target customer, and a big piece of the reason he started Luminous was because he didn’t have the hardware he needed. He’s spent a huge amount of time interacting with the end customers for our system, and understanding their real pain points. It comes through in our end-customer interactions, because they really love what we’re building. All we have to do is deliver it!
This customer focus is something that a lot of technologists miss; it’s easy to be a solution seeking a problem, especially the way that university research is structured here in the United States. Really spending the time and the effort to understand customer needs is crucial in every startup.
One of the hardest things, especially when the company is very small, is to differentiate between what is urgent and what is important. They're very rarely the same things. Carving out time to do both the urgent and the important is often impossible, because the urgent stuff requires dealing with interrupts, while the important stuff requires long-term concentration. This is one of the reasons that founding teams are so important; being able to divide the world between one person who deals with the interrupts and one person who can concentrate on the long-term stuff is often crucial early on. Defining roles and responsibilities early and having a conscious conversation over time about how they evolve, is often difficult but is really important.
Looking back, a common collaborative thread throughout your career appears to be a working relationship with physicist Tom Baehr-Jones, who serves as the current vice president of mixed signal at Luminous. What draws you toward this collaboration and keeps your professional relationship one worth continuing?
Tom is just one of the finest human beings I know. In addition to caring deeply about the people around him, he's also outrageously brilliant and is probably the hardest-working engineer I know. He’s got a real cross-disciplinary mindset that lets him see things that other people don't see, and he puts in the work to make them real. He's an incredible force-multiplier for any team that he's on; in every organization where we've worked together, he's built tools and systems that make the people around him 10x as effective as they'd be otherwise.
And he’s only one of the incredible engineers on this team. There are a really surprising number of absolutely amazing engineers on this team. It’s a privilege to work with these folks. My job is to make sure that the efforts of these brilliant engineers are going in the right direction, and clearing roadblocks for them when they show up.
As an established leader and founder of several successful companies, what is most important to you — a company’s mission, its vision, or its values?
That's like asking which of the three legs on a tripod is most important! You need all three, or the organization doesn't work right. And the three need to be aligned. If the mission is pulling one way and the values are pulling in another direction, you're going to get chaos.
Luminous Computing was founded by CEO Marcus Gomez and CTO Mitchell Nahmias in 2018. Since then, it’s evolved. Tell me about Luminous and its product-market fit. How do you achieve that fit and maintain it going forward?
The most important thing is to stay in very close communication with the customers. There’s just no substitute for that. In the end, the customer is the arbiter of whether what you’ve built is valuable. You can build all the fancy technology in the world, but in the end the customer decides what they want and why.
One thing that I really admire about this team is the sheer pragmatism that I’ve seen: The company started around optical compute, which was the heart of Mitch’s doctoral work. And optical compute has turned into a very active academic subfield; the work he did is right at the center of that activity. But the moment that it became clear that the customers really wanted something different, because the optical computing stuff didn’t solve a truly impactful problem, about 18 months ago, the founders did the hard thing and pivoted the company into a different architecture for the product. The optical compute thing is a neat science project, but it doesn’t seem to solve a commercially important problem. What we’re building now is going to have an enormous impact on a whole series of different commercial applications.
The past four years have brought incredible growth, with the company adding 60 new employees in the past year alone. It also just secured $105 million in Series A funding, including contributions from Gigafund and Bill Gates. What do you think that level of growth and investment will yield over the next five to 10 years?
Our intent is to revolutionize the way that machine learning models are built and trained. It’s not enough to just have more computation: We need to make that computation is easy for the customer to use as possible. We’re going to enable the development of models that are two orders of magnitude larger than anything being used today. And we’re going to make those models radically easier to develop and program, because our machine is architected so that the programmer has access to the full resources of the machine, without worrying about bottlenecked communications between different areas of the computer.
Knowing Luminous is also projected to hit 200 employees by years’ end, how do you go about hiring and developing top talent?
Excellence attracts excellence. The best engineers want to work with people who are their peers. Having a world-changing project is crucial for pulling the best people onto the team. Having the resources to really execute on it, and a plan that’s plausible and gets to a product quickly, is essential.
We interview very, very carefully. We spend an inordinate amount of time on it, and we’re conscious about interviewing both for technical competence and for cultural fit. Both things matter. We look very closely at people’s track record of delivering first-rate results, and we spend a lot of effort to make sure that the people we’re bringing onto the team are aligned in terms of both mission and value with the team.
One of the fun things about a startup that’s growing quickly is that there are lots of ways to satisfy ambition. People who want to gain leadership experience can do that. People who want to do things that cross organizational boundaries can do that. There’s always more to do than people who can do it.
One thing I really love is to bring people onto the team who know how to do multiple things, across multiple domains. People with a track record of learning new stuff rapidly and then making a contribution at the intersection of two or more different technical stovepipes are incredibly valuable.
Looking over the entire landscape of AI, what do you think the biggest challenge will be for the industry in the immediate future?
AI is going to produce a lot of economic dislocation and a lot of societal disruption. A lot of things that used to be expensive will become free, or nearly free. New capabilities are going to show up where the computer can do things that used to be the exclusive domain of the human mind — and in many cases the computers will be better at them than any human ever was.
Every technological revolution in human history has produced economic and social consequences, many of which are very hard to predict in advance. It’s critical that everyone in this field think consciously about how to ensure that the tools we’re building are put into the right hands and deployed in a safe way. As a company building incredibly powerful tools, we think a lot about what hands those tools should go into, and I think that making those decisions is going to be a challenge for the whole field going forward.