Thursday, January 21, 2021

My In-Browser, Serendipity Hangout & Spatial Video Side Project: PLUTO!

I am excited to share Pluto, a side project I have been working on with my friends at Fluxon and Katsuya Noguchi. You can check it out here at

I got sick of all the Zoom happy hours, birthday parties, and company get-togethers. In parallel, fun random chance meetups largely died - you could no longer run into a friend or co-worker or interesting new person given social distancing[1].

Pluto is an online in-browser spatial video for meetups, serendipity, and video chat:

  • A browser-only video space (no headsets needed) you can move around using first-person shooter-like controls. 
  • Spatial video chat - Sound decays with distance so different groups can break up and hang out. This allows for hanging out with large groups in a way that feels like the offline world.
  • Just drop in a link and you can attend an event over video (no headset, no nothing). 
  • Everything is represented in 3D, which means people can randomly run into each other and serendipity can finally happen again. 
  • You can chat, watch videos, screen shares and explore virtual spaces.

Pluto is fun to explore on your own, but is actually even better with friends. 

We have held happy hours with 50-60 people at a time and it works magically - you can run into people, walk over to a group, and in general interact in group or side conversations in a seamless and organic way. It is the closest thing to a real world event, but online.

Pluto has many worlds to explore

We have built out a number of worlds you can explore in Pluto including:

Here is a picture of Elon Musk hanging out at Burning Man on Pluto. He has not been on Pluto as far as we know, but mocks allow us to pretend. Come check out Burning man and the video screens with DJs!
  • The Maldives. Relax. Enjoy the ocean. Finally a break from your basement zoom call!

There are some easter eggs hidden in this library - can you find them?

Come check it out and let us all know what you think! We will be hanging out at an instance at from 11am to 2pm PT. Come and say hi!

You can also grab your own version at You can explore it alone, but is actually more fun with friends. Try inviting 5 to 50 people and see how it goes!

Please let me know what you think, comment on Hacker News, or ping me on twitter.

It is still early days, but thought it would be fun to get it out there and get some feedback :)


[1] The rise of webRTC and webGL has enabled a new, real time, in-browser social layer to emerge. No one has really used this fully yet, although apps like Figma obviously are webGL heavy for collaboration while social products and experiments like Clubhouse, Gatherly, Rally, Rambly, Sonar, Teamflow, Yorb are doing interesting things (and helped inspire Pluto). So, we put together Pluto - built on webRTC/MediaSoup, and THREE.js.

Thursday, January 14, 2021

Why Has Israel Succeeded At COVID Vaccination?

Israel has pulled ahead of much in the world in its rate of vaccinating its citizens - with roughly 1% of the entire population vaccinated per day and over 23% of the country vaccinated in the first few weeks. The country hopes to have the entire population over age 16 vaccinated for COVID by end of March.

Israel focused on first vaccinating people by age group (and then comorbidity) with the idea that if you vaccinate the 20% of the country that represent 95% of the deaths, you can avoid deaths from COVID.

% of Israeli population within a certain age vaccinated by 1/13/21. Chart via Segal Eran

Early data is starting to suggest Israel’s approach is working. For example, in a cohort of 200,000 people vaccinated in Israel, there was a 12X drop in COVID cases by day15-22 of a single dose of Pfizer/BioNTech vaccine.

The right hand column below is day since first shot to result. The left hand side is # positive for COVID.

In parallel much of the rest of the world has been slower on rolling out vaccination with UAE as an exception. 

After speaking to a number of members of the Israeli public health community involved with COVID, the following picture emerges on why Israel has been so successful:

1. Focus on simplicity & pragmatism. 

The biggest takeaway is a focus on simplicity and pragmatism. For example, the criteria for vaccination are simple - you are either a healthcare worker (including nursing homes & assisted living facilities), or you are tiered by age (and later comorbidity). 

Healthcare workers have been shown to spread COVID and other diseases within hospitals (known as “nosocomial transmission” and a key aspect of SARS and MERS - two other coronavirus based diseases), while older people are most vulnerable to hospitalization and death from COVID.

Having age-based tiers makes it easy to know who should show up and who is eligible, and removes a big burden on enforcement. 

This simplicity and pragmatism extends further. For example, in the USA the roll out of vaccination in California to healthcare workers was slowed by complex tiering within healthcare workers themselves based on how much patient facing time each type of workers has.

In contrast, the Israelis I spoke to said “We vaccinate an entire workplace. We assume everyone will eventually be vaccinated so we found it easier to just show up and vaccinate every single person who works at a hospital. We do not care if they are doctors or administrators or whatever - it is easier to just vaccinate everyone”. 

Many countries and states have been too focused on “fairness” and “equity” so have frozen their vaccination efforts in place, or put in place large fines for “misused virus”. Remember - everyone will eventually get vaccinated. The more shots in arms, the better, with an emphasis on the old and comorbid. And also remember, we are in the middle of a “once in a century pandemic”- it is more important to move fast to save lives than to create and enforce complex rules. However, it turns out these complex rules are not needed - there are simple criteria for who will get sick and die of COVID per below.

2. Vaccinate the people who will die.

The biggest risk factor for death and hospitalization from COVID is age. Age outweighs comorbidities in some cases by as much as 30-fold. A handful of comorbidities matter, but age dominates.

Simple math suggests that vaccinating roughly 20-25% of the population would prevent 95% of COVID deaths in many western countries. So, pragmatically, the Israelis focused on vaccinating that high-risk 20% of their population first. This has mapped to age-based tiers (and a handful of comorbidities) which simplify vaccination roll outs. They also vaccinated healthcare workers and long term care facilities first.

Many of the calls in the US for “ethics and equity” in vaccination seem to be politically motivated versus science based. Complex, multi-tier criteria that prioritize young “essential” workers over 70-year olds, who are much higher risk of dying of COVID, means more people will die as we wait to get to the 20-25% of the population who actually matter in terms of risk of hospitalization and death. There is little logic in vaccinating 30-50% of the population (depending on your definition of "essential workers") and then 20% of high risk people (50-70% of the entire population between them!!) to protect 20% of the population who contribute 95% of COVID deaths. It is much easier to focus on the 20-25% who are actually at risk first. It seems unlikely it is ethical, or equitable, to let people die for your “ethics & equity framework”.

The science is clear - vaccinate the elderly, then people aged 16+ with certain comorbidities and you will see COVID deaths plummet. (Side note, the CDC finally updated their criteria to 65+ while this document was being written). We should all honor the role essential workers have played in keeping the country and world open. One way to honor this is to protect their elderly and sick family members from dying of COVID - by vaccinating their at-risk family members first. After the people with highest risk of death are vaccinated, the subset of essential workers who did not fall under those demographics can be vaccinated next. For example, Israel started this week vaccinating 55-year olds and up, as well as all teachers.

The nice aspect of this approach is you do not need to vaccinate that large of a population before you see a big impact and can potentially reopen. If fewer people will die of COVID (once vaccinated) then other common diseases like flu going forward, a country or state can reopen with many lives saved.

Here is Israel’s prediction on what to expect in terms of deaths:

3. Create as many endpoints to vaccinate out of as possible.

The Israelis I spoke to emphasized their focus on opening as many vaccination clinics and centers as possible. They mentioned that when a hospital set up a vaccination program, every possible clinic in the hospital was also opened for vaccination.

Similarly, parks, schools, and other public spaces have been converted into places people can go to be vaccinated.

Israel has also opened some “mega centers” for vaccination, per the picture below.

Some cities like San Diego have recently followed suit after feeling stymied by State and Federal government:

4. Don’t waste vaccine.

Remember, everyone will eventually be vaccinated. Throwing a scarce vaccine in the trash is an enormous waste. Rather than waste vaccines, the Israelis have two mechanisms for overflow. First, anyone can wait in line by a vaccination center starting at 7pm and if there is vaccine left over, be vaccinated. Second, if no one is waiting in line nurses or other vaccinators will go out into the street looking for people to vaccinate (and then schedule their next appointment for the second dose on the spot). There is a famous story of nurses coming out of a clinic and spotting a pizza delivery person. “Hey pizza guy want a vaccine!!” they yelled to call him over and then vaccinated him.

5. Success begets success.

Israel has been incredibly transparent on data around the vaccination program - with everything from dashboards showing vaccines used per day, % of population by age vaccinated etc. to a Telegram channel from the Ministry of Health with daily data dumps to whomever signs up.

One general tenet of life is “success begets success”. By vaccinating rapidly the pharma companies approved in Israel (Pfizer and Moderna) have continued to prioritize the company for more vaccine as capacity comes online. While Israel originally expected a gap in deliveries and temporarily running out of vaccine in January, that gap has now been bridged via new pharma deals.

The US could consider doing something similar - allocate more vaccines to the states that are actually using it. Hold back vaccine from states that are not moving quickly until they move fast enough to use it.

Remember - we do not need to vaccinate everyone to have a big effect. If we vaccinate 20% of the population it might be possible to still drop deaths 95%, which is the primary goal of the COVID shelter-in-place, social distancing, and vaccination efforts.

Factors that are mentioned but seem overweighted (AKA excuses).

There are a lot of reasons (perhaps a better word is “excuses”) that are made for why Israel is succeeding while others are not. This includes things like:

a. Small population and geography. 

Israel is a country of 9 million people. The argument is its small size makes it easier to vaccinate. The reality is that many countries the same size or half the size of Israel are doing a much worse job of vaccinating their populations including Denmark (5 million), Norway (5.3 million), Netherlands (18 million) and others. Similarly, States like New Jersey (population of ~9 million) are far behind Israel as well. If size were the only constraint, we would see more success elsewhere.

b. HMO/healthcare centralization.

The Israeli population is largely covered by 4 HMOs - with the largest, Clalit, covering roughly 50% of the population. Centralization of healthcare services undoubtedly matters in decision making speed and administration. However, as reminder the fragmented US system vaccinates 50% of its entire population age 2 and up for flu every single year - and does so in a matter of weeks. The primary obstacles in the US seem to have less to do with centralization and more with complex criteria and fears of scarcity of vaccines driving even more scarcity.

c . It must be the Israeli military coordinating all this!

The Israeli military is not coordinating the population’s vaccination. It is driven by a combination of the Ministry of Health, the HMOs, and some private companies that have been contracted for a subset of the logistics. This is all do-able in the USA too.

d. Israel is a wealthy country! That must be why! Or they spend more on healthcare!

Israel GDP per person is lower than the USA overall as well as lower than states of the same size like New Jersey. The country also spends around 7% of GDP annually on healthcare - roughly one-third what the USA spends.

e. The culture is different!

Israel has more in common with the US than one might guess. There is a fractured, contentious political system. A number of healthcare workers were going to refuse the vaccine - until they were told they would be placed on administrative leave if they refused it. A meaningful subset of the population ignored their second lockdown.

The USA can do it too!

Every year the United States vaccinates 50% of its entire population over the course of a few weeks for the flu. In past outbreaks the country has also been fast to move. For example, in 1947 the city of New York vaccinated 5 million people in 2 weeks to combat a smallpox outbreak.

If needed, the country can move fast. In order to do so, it should:

(1) Simplify criteria. Make vaccination age-based (and include healthcare workers and assisted living facilities). Be pragmatic. Vaccinate the whole hospital versus quibbling on the order in which staff get it. Remember - everyone will eventually be vaccinated. The two goals are to stop deaths and then to build herd immunity - in that order.

(2) Focus on the people who will actually die of COVID, not politics. 20-25% of the population vaccinated may drop deaths 95%. The CDC is finally moving in this direction, but should focus on age-based and then co-morbidity based tiering versus other factors. “Fairness” falls out naturally if everyone is vaccinated in the next few months and everyone’s family members’ lives are saved, versus taking an entire year with incremental disease and deaths.

(3) Open as many vaccination points as possible as fast as possible

A simple model is that anywhere that administered the flu vaccine can be used to vaccinate for COVID. Or, augment the pop-up testing centers with vaccination services. These stations can be staffed by healthcare workers from the clinics being used, as well as pharmacists (trained in flu vaccination), dentists and others.

(4) Don’t waste vaccine. If there is excess left over, use it on anyone so we can build herd immunity faster. 

(5) Optional - reward success. If a city,region, or state moves extra fast at vaccination - give them more. Reward the communities that vaccinate quickly and create competition and a leaderboard to get it done.

We finally have a vaccine. Many people are now dying unnecessarily due to a lack of delivery. The Israel model suggests a clear way to move forward. Let’s get it done.

Friday, December 18, 2020

Index Companies

Sometimes there are markets that are clearly going to grow massively over time. For example, ecommerce, genomics, crypto are all markets which were clearly going to compound over time. It might have been tough to call the winners of each market early, but it was clear the markets themselves would grow a lot.

One lens through which to view companies is to ask "what companies are an index of their underlying market"? 

Index companies often take a cut of every transaction in their space, or are a piece of infrastructure everyone in the market needs. For example, it is a hard to launch something into space in the West without using SpaceX. These companies may be ways to participate in the market broadly without having to worry about how wins in it.

Some example index companies are:

Online Commerce & Stripe. 

For many years Stripe effectively acted as an index of startups in online commerce. One can argue Stripe is now a proxy for the underlying global online economy. Buying Stripe means buying an important subset of the internet.

Crypto & Coinbase. 

Crypto is a dynamic market with hundreds of tokens beyond BTC and ETH. As long as Coinbase lists it and takes a cut of transactions, Coinbase is a proxy index for all of crypto. With its pending IPO, Coinbase will effectively be an index for crypto writ large. Coinbase is also unique as a high quality company which institutional investors can use as a proxy to participate in the crypto world, even if they are not allowed to hold crypto directly themselves.

Genomics & Illumina. 

I would not be surprised if 90% of clinical genomics were run on Illumina machines. Illumina stock is a proxy for the clinical use of genomics.

Space Launch & SpaceX. 

Today a majority of commercial space launches for the western world are conducted on a small number of companies with SpaceX as the majority of that share. There is a reasonable scenario that in 5-10 years time most of space launches will consolidate to SpaceX and Relativity. Right now SpaceX is a proxy for the entire space launch market. Relativity seems like a strong #2 technologically, although yet to launch a commercial rocket.

AI & Chips Companies. 

The best way to index AI is likely through an underlying AI-chip company. Today that is NVIDIA but GPUs are not really optimized for ML applications. One could argue Google should have launched its TPUs for others to buy directly, versus as a cloud service. Since it has not, potential contenders include Groq and Cerebras.

Fragmented indices

Some markets have a few players that index them - or the revenue mix of the companies prevents a pure play view. For example, AWS, Azure, and GCP are an index of the cloud, but none are investable as stand-alones.

Future indices?

It is fun to speculate on what markets or companies may have index companies emerge. For example, is Substack an index of next-gen journalism?  Zoom an index for remote work? 

Or you can turn it around and ask "What trends are important? Are there companies that represent a proxy for all the activity in this trend?" For example, what is an index for 3D printing? Or DeFi (is it ETH?)? Or....?

What are other "index companies" for exciting markets? Let me know on Twitter.

Friday, October 23, 2020

Unicorn Market Cap & Industry Towns, 2020

In 2019 I wrote about how "industry towns" emerge in every market. These clusters of people, ideas, capital, service providers, and companies tend to have strong network effects that support startup formation and success in a given industry. 

For example, Silicon Valley, London, Beijing are global tech centers. New York, London, HK, Shanghai etc. are global finance centers. Hollywood, Lagos, Bombay are global movie centers.

Given all the IPOs in the last 16 months (including Snowflake, Unity, Asana, Palantir) and the sharp decrease in market cap in some companies (WeWork), I thought it would be interesting to update the view of where unicorn market cap currently resides. 

Some big caveats include (1) Unicorn market cap is a lagging indicator of ecosystem health since many companies take anywhere from 2-7 years to be worth their first billion (2) COVID and the move to more remote work or "remote first" startups may impact what this looks like in 4-5 years (3) San Francisco governance may decrease San Francisco's long term relevance, although the broader Bay Area should be strong longer term. A big thanks to the talented Shin Kim for pulling & structuring the data below (raw data here if you want it).

Unicorn Turnover
Since June 2019, 37 of the 361 Unicorns at the time went public, 14 were acquired, and 8 did down rounds or shuttered. 187 new unicorns emerged in the last 15 months (>3 a week!) which means 38% of total unicorns today are new ones (and the number of unicorns grew 58% in 15 months!).

Half of the 187 new unicorns since June 2019 are in 5 cities
88 of the 187 new unicorns since June 2019 are in Silicon Valley, New York, Los Angeles in the USA and Beijing and Shanghai in China. Roughly 25% of new unicorns globally were in Silicon Valley. This is inline with Silicon Valley's traditional share of the overall unicorn market.

There were new unicorns in 65 different cities.

Below is the distribution of # of new unicorns. Only 8 cities in the entire world added 5 or more unicorns since June 2019. Only 4 cities added 10 or more unicorns - add accounted for 43% of all the new unicorns in the world over the last 16 months.

Overall Unicorn Market Cap: US + China = 77% of Unicorn Market Cap
As anticipated by industry clusters and network effects, over 3/4 of all unicorn market cap is in 2 countries, with high regional concentration within countries.

The USA and China continue to lead in terms of both Unicorn market cap and number. Of China's $512 billion in Unicorn market cap, almost 40% of it is from two companies - ByteDance ($140 billion market cap) and Didi ($62 billion).

Decacorns, or companies with $10B or more in market cap, are similarly concentrated in the USA and China representing 19 out of 26 total global decacorns.

In-country concentration
Just as unicorn market cap and number are concentrated by country globally, each country also has its regional tech clusters or industry towns.

In every country surveyed almost half or more of all unicorn market cap, and number, resided in a single city or major metro area.

Unicorns in the USA

In the USA, Silicon Valley is roughly 5X the size of New York and Los Angeles by market cap (of which 2/3 of the market cap is Space). Boston Unicorn market cap is biotech driven. SV is 5X NY in terms of # of unicorns, and ~8X the size of Los Angeles.

Next tech cluster in the USA
Intriguingly, a number of the cities often spoken about as the "next tech cluster" have not added much by way of unicorns. Austin has 2 total and Portland has 1. One can argue that this is simply a time lag and in 3-4 years many unicorns will have sprouted locally. Anecdotally, the cities that seem have a number of high quality founders and new companies - and therefore may be on their way to growing more unicorns include Denver and Salt Lake City. While I see many investors move to Austin, I see more founders move to Denver. Time will tell.

China Unicorns
In China, Beijing is 7X the size of Shanghai by market cap but only 2X by unicorn number. This is driven largely by ByteDance and Didi. Subtracting these two companies off yields only a 3X in market cap difference between Shanghai and Beijing. It is notable that both Beijing and Shanghai added 11 unicorns in the last 15 months. So new Unicorn growth is more neck and neck more recently.

India Unicorns
India presents an interesting case study in a second cluster forming. When I visited India in 2007, Bangalore was the clear engineering center and tech startup leader. Hyderabad was a strong tech customer support center. Since that time New Delhi as emerged as a major tech cluster and startup unicorn hub. While Bangalore continues to thrive, some point to bad governance locally (traffic, high rents, passive government) as one of the reasons it is not the only center. There are undoubtedly other reasons, but it is worth thinking about relative to San Francisco longer term.

Other markets
The pattern of a primary tech and unicorn and tech cluster repeats itself for the other major markets including Israel (Tel Aviv), Germany (Berlin & Munich more neck-in-neck), UK (London) etc.

Outlier market cap domination
Startup market cap tends to follow a power law. In some markets, there is very high market cap concentration currently (this will of course change when the companies go public). For example, 58% of Beijing's market cap is in two companies - ByteDance ($140 billion) and Didi ($62 billion). Los Angeles unicorn market cap is largely a function of SpaceX (2/3 of the total!). Silicon Valley market cap is more distributed, perhaps suggested a broader tech ecosystem.

Why are all the unicorn "Remote First" companies since COVID in the Bay Area?
Out of the top 100 private tech unicorns, 11 have perpetual "remote first" or "remote equal" policies. 5 of these were added since COVID. Interestingly, 4 of 5 (AffirmBrexCoinbaseFigma) of the unicorns who announced moving to remote-first forever post-COVID are all based in San Francisco, with 1 (Quora) based elsewhere in the Bay Area. 

Given that roughly half the unicorns in the USA are based in the Bay Area, you would expect roughly half the unicorns going remote first to be based elsewhere. Instead all of them are in Silicon Valley (roughly a 1/32 chance if this were to happen randomly). Obviously, something else may be afoot.

San Francisco should be going through a golden era and boom. It could be one of the great cities of the world, with a large city budget to support it. Unfortunately, San Francisco's poor governance, high cost of living, lack of housing, failed homeless services, bad budget controls, and other issues were prominent before COVID. With a never-lifted lockdown the city has continued to degrade and this is reflected in both the homeless situation and dropping rent prices as people move out of the city.

The smoke and fires was a recent last straw that drove a set of young unmarried founders I know out of the Bay Area. Anecdotally these founders have largely temporarily relocated to New York and Los Angeles. The main reason people stay in SF is the social network and the serendipity of a cluster and network effect. Due to SIP, if you can not meet people in person there is no impetus to put up with poor living conditions and to stay put. Many have decided to go to cities that are more liveable, with less restrictive conditions.

In parallel, a number of later stage startup CEOs are interested in hiring more outside of the Bay Area, and especially outside of San Francisco. There is an increasing perception that San Francisco based hires are more expensive, self-entitled, and political-activism-at-work oriented (versus core company mission & performance focused) than employees elsewhere. Of course, not all late stage founders share this perspective but an accelerating number do. This impacts early stage companies less with their smaller employee bases, tightly defined missions, and existential need to ship. 

If the Bay Area does eventually decrease as a cluster, I am guessing it will be only a partial decline. The network effects and talent base in the Bay Area are incredibly high and if diminished I would guess would drop by 10-20% not 80%. That said, New York, Los Angeles, and to a lessor extent Denver, seem positioned to grow in importance over time if COVID SIP extends indefinitely or SF takes an aggressive tax stance or continues on its governance slide. While I am long term bullish on the Bay Area as an ongoing home for unicorns and major US cluster, it is worth watching how ongoing governance and policies will impact SF. Only time will tell. 


  • Unicorn list and valuation as of 10/7/20 from CB Insights
  • HQ location based on company websites and Crunchbase

You can order the High Growth Handbook here.



Startup life: