Monday, October 11, 2021

MegaCycles in Tech & Crypto

Every 8-10 years, the technology industry used to go through a boom and bust cycle. A new technology or platform would emerge, there would be rampant investment and speculation, a few strong hypergrowth survivors would emerge and most of the rest of the new startups would collapse or get consolidated. This happened with semiconductors in the 60s & 70s, microcomputers in the 80s, and the internet in the 90s. 

Each successive wave was bigger than the prior - both in terms of market cap created as well as money that flooded in. 

In the 2000s until now, something odd happened. The venture boom and bust cycle stopped. Things in tech have been replaced by a single long ~20 year boom. Even the financial crisis of 2007-2009 did not impact tech startup formation or growth much.

While there are undoubtedly multiple drivers for this lack of tech cyclicality [1] one potential explanation is the stacking of technology waves on top of each other. Instead of a single cycle drawn out over 8-10 years for a single new technology, we are now seeing multiple overlapping tech waves all happening on top of one another. This is both increasing the size of the overall resultant result, as well as smoothing out any down cycles.

For example, from 2005 until now we have had overlapping waves of cloud, social, mobile, SaaS, vertical SaaS, fintech, AI, and crypto. All of these would have had their own 10-year cycle in the past. One could argue we are now going through a mega-cycle ("poly-cycle"?) which may last for (at least? at most?) a few more years. Part of this cycle is dependent on capital availability and quantitative easing, but a lot of it is just software eating multiple industries simultaneously. COVID was a big driver of tech adoption as well across both consumers (Instacart, DoorDash, Amazon, etc) and enterprises (Zoom, Stripe, Figma).

One of the main characteristics of a megacycle is the lack of downturns. Instead of a sharp recession in tech leading to lots of layoffs, companies dying etc, the good times just keep rolling as wave after wave of new technology shifts overlap. This has both good and bad side effects to be discussed in a future post.

The time between these cycles has been collapsing, and the size of each cycle increasing. As each wave accelerates, it also may accelerate subsequent waves. For example the 10Xing of people online, time spent online, spend online, have all accelerated each other and their underlying technology waves. Sometimes adoption of something increases more adoption due to network or scale effects, versus slows things down (although large numbers inevitably catch up).

Interestingly, crypto itself has previously had much steeper cycles on roughly 4 year cycles (timed with Bitcoin halving and therefore a sudden shift in supply/demand in crypto leading to bitcoin and then alt-coin runs). The first crypto cycle was effectively the bitcoin white paper drop + initial mining. The second cycle was the emergence of Ethereum, ICOs, and new protocols & tokens. Many crypto people I know assumed that in 2021 we should have had a down cycle in crypto.

One potential explanation for a lack of downside is the massive money printing and spending being done by the US government, which is now literally just sending people cash. This creates both monetary stimulus that will inflate multiple assets, as well as potentially have investors seek crypto as an inflationary/low of dollar value hedge.

An alternative, or overlapping explanation one could argue is around overlapping crypto cycles - or a crypto mega-cycle. Just as tech had multiple overlapping waves, crypto is now seeing the same thing with multiple overlapping waves of DeFi (Uniswap, Aave, Maker, etc.), NFTs, DAOs, and the launch of multiple interesting layer 1 protocols all started a few years ago (Solana, Near, Celo, Minna, etc.). There are also an increasing number of efforts being built between crypto and traditional fiat rails, as well as broader adoption trends. Given the diversity of crypto efforts, a sudden drop and decimation in one part of the market (for example if NFTs hit a bump for some reason) may be smoothed out by a new trend or rise in another (DeFi or BTC running). In other words, parts of the market may short-term decrease their correlation a little over time as the footprint of crypto and its uses cases expands.

Some argue Crypto may also have gotten large enough in terms of usage and market cap that the extremely large fluctuations from the past may dampen a bit. Loosing 50% of a $2.4 trillion market cap is a $1.2 trillion shift - but you still have over $1 trillion of market cap left and potentially lots of buyers in the wings due to sheer scale and multiple use cases. 

Sustainability of market cap may reenforce the reflexive nature of crypto. Even when crypto dropped 50% in 2021 it still had a $1 trillion+ market cap and it was clear it was not going away for good. This drives more participants ongoing into the market so there will be more buyers and users and the potential for a stronger snap back. Similarly, great talent may continue to enter the market in the absence of a sharp dip. In 2000-2001, there we mass layoffs in technology and many people who left tech did not come back. In 2017 talent into crypto slowed as the downcycle hit. Currently the better sustainability of crypto market cap means more tech and new grad talent continues to come into crypto which should help push the next innovation waves in the market.

It will be interesting to watch the coming year as to whether crypto goes back into cyclical behavior with a sharper downturn soon, or if we are now caught in a megacycle. 

The big confounder on this all is ongoing mass scale government money printing and spending. Given ongoing supply chain and energy issues and pricing rising in various industries, more people seem worried about the chances for transitory or prolonged inflation or "stagflation"[2]. Bitcoin was started in part as a reaction to the great financial crisis and US monetary policy. One could argue the early origins of cryptocurrencies was to hedge the exact environment we now find ourselves in[3].

Thanks to Matt HuangFred Ehrsam, and Curtis Spencer for comments on this post.


[1] For example, one aspect or cause of the boom may be explained by a few things including the ongoing need for capital to find a high return / high growth home due to low to no interest rates caused by quantitative easing. So a lot more money has flown into tech over time due to a lack of yield anywhere else.

[2] Stagflation is when there is rising inflation - so companies need to pass on higher pricing to cushion margins. However those higher prices result in demand destruction and a stagnant economy.

[3] So "hedge US macro" may be sufficient cause for a flight to crypto right now and the quick rebound from a down cycle.


Old Crypto Stuff:


Startup life

Sunday, September 26, 2021

The False Narrative Around Theranos

One of the interesting aspects of the Theranos trial is the degree to which some folks are buying into part of the defense's narrative that "Theranos was just acting like every Silicon Valley startup". This is of course blatantly false, but it is being adopted as some form of truth. 

The claim is that every tech founder somehow pushes the envelope of truth, and therefore that is all Theranos did (versus potentially committing fraud over a 15 year period, lying to regulators, physicians, employees and investors while endangering patients).

This analogy breaks down on multiple levels. 

There is a big difference between drunk driving at 90 mph in a school zone versus driving 5 miles too fast on the freeway. (Or, in the case of most tech companies, simply respecting the speed limit).

While there are obviously some bad actors in tech, there does not appear to be quantitative evidence to suggest this is any worse than in non-profits, finance, mediahollywood, or any other sectors

Where the analogy breaks down between Theranos and the "average" tech company:

1. Most technology companies do not lie about their product or service. If they did so, they would not be able to attract or retain great employees, or scale revenue and product adoption so rapidly. If their product did not work, no one would use it. This is especially true over a decade+ journey. Despite some high profile counter-examples, most tech companies are honest companies run by people who want to do good.

2. Over its 15 year history (and $700 million raised), Theranos never had a working product. It appears possible Theranos' approach was potentially unlikely to actually be able to work based on chemistry/biological contamination of approach. Think about this for a minute. Theranos apparently lied to people about its product's potential for 15 years without ever making it work. 

Additionally, given the small titers of blood, the finger prick as a source of contaminant versus venus blood draws, and other aspects of the chemistry, some believe the Theranos approach is extremely hard to make work from a chemistry perspective.

3. The company launched a fake product to living, breathing patients whose potential course of treatment and therefore life and death situations depended on accurate results. This is different from a telling a CIO that your data science tool for their customer support team would be ready in Q1 and missing the deadline. Real harm happened to real people, who course of care depended on Theranos results.

4. The scale of lying was exceptional. Theranos appears to have misled regulators, employees, investors, partners, physicians and patients. It immersed itself in secrecy, even internally, to be able to keep employees from talking to each other and for the ongoing deceit to be detected. This is different from most tech startups where transparency is often one of its early core principles or approaches. From weekly all-hands to internal Looms, tech startups tend to be highly transparent places to work.

5. Theranos raised no real mainstream venture capital. None of the mainstream tech or biotech funders invested in Theranos. Given that there are tens of reasonably good venture firms, this is striking.

6. Theranos was a diagnostics company, not a "tech" company. It is striking how many non-tech companies that blow up did not really ever have much to do with technology. Theranos fit squarely in the "medical devices & diagnostics" world and its focus and (never-quite-worked) innovation was on the chemistry and hardware for biology side. WeWork - another company often pointed to as a "tech company", which had issues for other reasons (but started off as a viable business) was a real estate company. Branding one's own company as "tech" tends to be give it higher multiples, access to more money, and a brand allure with media. Eventually reality tends to catch up.

An increasing portion of the discourse in the USA today seems to be anti-tech, anti-maker, and anti-success. This stance probably reflects more of what is currently happening in the people writing these negative pieces (or opining in articles) rather than in the tech industry itself. Theranos is being used as a catchall example to drive this false narrative that people can not do good work, benefit millions of people, and make money, without somehow being nefarious. I encourage you to not buy into this false hype. :)

Monday, June 21, 2021

Unicorn Market Cap, June 2021 (Almost Post-Pandemic Edition)

I have previously written about Unicorn Market Cap and Industry towns in 2019 and 2020. Over the last 8 months the number of tech startups worth $1B or more ("unicorns") has grown by 43% from 487 Unicorns to 701. This is almost double the 361 unicorns in June 2019 (!). 

Data was taken from CB Insights and a special thank you to Shin Kim, CEO of Eraser for the data and graphs. 

Caveat emptor: data from CBI is updated/reconciled over time, so very recent unicorns may not be included yet. However this provides a directional view.... Raw data here.


The regional nature of private tech market cap continues to dominate. The big shifts over the last year include:

(1) United states: Over 67% of the new unicorns by # are in the USA with 154 total. (out of 227 globally)
There were 69 new unicorns in Silicon Valley, 30 in New York, and 8 in Los Angeles. 

This is increasing share for both the USA & Silicon Valley as a % of global tech unicorns, with NY and LA accelerating somewhat. New York anecdotally feels like it has transitioned into a break out cluster of its own.

The pandemic has increased unicorns in the USA at a fast clip.

(2) China has slowed on new unicorn generation.

While China is 29% of all unicorn market cap, it only added 9 new unicorns (roughly 4% of global total) since October 2020.

The decline in new unicorn formation in China is striking. One potential interpretation is it at least in part a data issue. For example, 20 or so Chinese unicorns from pre-2020 were just added to this data set as a historical reconciliation. Other interpretation in the last section below.

(3) Europe added 25 unicorns.
London (8 new unicorns for a total of 21), Paris (5 new unicorns for a total of 13), Berlin (5 new unicorns for a total of 9), and Stockholm (2 new unicorns for a total of 4, including a decacorn) added the most new unicorns. 

(4) India added 11 unicorns.
Major cities to add unicorns included Bangalore (7 new unicorns for a total of 14), New Delhi (2 new for a total of 12), Mumbai (1 new unicorn for a total of 4), and Chennai (1).  


The overall Unicorn rankings have remained the same. The USA is the clear front-runner, China next, followed by the EU and then India. Israel continues to have a large number of unicorns per capita and Canada has started adding them at a faster clip then before (4 new unicorns in Toronto alone in the last 10 months). Brazil is the biggest generator of unicorns in Latin America.

Unicorn market cap is highly concentrated by specific cities:

DECACORNS: $10B or larger market caps
Decacorns are largely concentrated in a handful of countries - although a number of decacorns have gone public in e.g. Indonesia and others.

Within each country, specific cities or regions continue to make up 50% or more of the country unicorn market cap.

For example, Silicon Valley is 51% of the US market cap and 47% of unicorns by number. New York is 11% of market cap and 17% of unicorn by number, and Los Angeles is 11% of market cap (largely due to SpaceX) and 7% of US unicorns by #.

Clusters & industry towns clearly continue to matter.

Europe has a number of centers with London the largest by a margin, followed by Stockholm, Paris, and Berlin.

Beijing (59% of market cap due to ByteDance and 37% of # of Chinese unicorns) and Shanghai (13% of market cap and 24% of unicorns) are the largest in China.

While 77 cities around the world have at least 1 unicorn, most are concentrated in 13 cities with 11 or more unicorns.

"New tech clusters" are not big clusters yet
A lot of great marketing has occurred for Austin and Miami during the pandemic. Austin added 3 unicorns to get to 5 total (below new additions of unicorns in SV, NY, LA, Chicago and Boston and tied with Seattle, DC, Philadelphia). Miami added 1 unicorn to get to a total of 3 (although I am aware of at least one other that just moved there from LA that does not appear counted in the data yet). I am anecdotally bullish on the long term prospects of Miami given the people I know who moved there and the frontier feeling it has. Under discussed as future clusters with strong prospects are Denver (2 new unicorns for a total of 4) & Boulder (1 new unicorn) and Salt Lake City (2 new unicorns for a total of 4). 

As noted above, Silicon Valley continues to make up 51% of market cap and 47% of total unicorns in the USA. NY has grown its share of unicorns by number relative to other US cities.

Here is the full list for the USA

Beijing, Shanghai, and to a lessor extent Shenzen continue to drive unicorn # and market cap in China.

In India Bangalore and New Delhi continue to run neck in neck, with Bangalore having 2 more unicorns.

EUROPE (and Israel)
Europe continues to see strong clusters in the UK (where London easily dominates), Germany (ditto for Berlin) and France (ditto for Paris).



In Israel, Tel Aviv continues to be the main cluster.

In parallel to new unicorn formation, unicorns also "graduate" via an IPO/DL/SPAC, an acquisition, or going out of business
As unicorns graduate the founders and employees of the companies may either fund other new ventures as angels or VCs, or start new companies themselves. It will be exciting to watch more tech ecosystems grow as more unicorns go public.

This data should be caveated as new unicorns from past years are sometimes added by CB Insights (who have done a great public service by aggregating this data to begin with!).

So caveat emptor on over interpreting this data. However, the difference between the USA and China on new tech unicorns over the last 2 years in notable.

Possible interpretation would include:
1. Data fidelity. Is China reporting delayed?

2. Keeping it quiet. Given the unusually high death rate (a Chinese billionaire dies every 40 days) and kidnappings of Chinese billionaires, perhaps there is now a disincentive to announce highly valued financing rounds? 眼不见,心为静。 眼不见心不烦。

3. China is producing fewer tech unicorns. This seems odd given the acceleration seen in tech due to the pandemic, but is possible. 

Some of the top countries for rest of world look like this:

Raw data:



Startup life

Thursday, June 17, 2021

Anduril & Defense Tech

The last year has demonstrated repeatedly the lack of societal preparation for multiple forms of threats to our country and world. Examples of this include issues responding to the COVID pandemic, the cybersecurity ransomware attacks on critical US infrastructure such as our energy systems and food supply, and of course the ongoing issue of climate change[1]. In parallel, new technologies are increasingly being used by antagonistic groups and countries for terror and war. Drone attacks on international oil pipelines and their use in regional conflicts has increased dramatically in the last few years, as have drone related incidents and accidents around airports, sports stadiums, and infrastructure such as power plants. 

Just as old incumbent institutions with little to no organizational renewal impacted our ability to respond to COVID, the defense industry has undergone significant consolidation over the last 30 years. There has not been a new defense technology company of any scale to directly challenge these incumbents in many decades (SpaceX and Palantir are obviously sizeable ~20-year old “newer” entrants in adjacent areas).

One exciting newer defense technology company that is working on building capabilities around sensor-based awareness and anti-drone activities is Anduril. I am excited to lead their latest round of funding. Anduril will use this funding in part to drive new acquisitions, as well as an ongoing ramp in their team and business. Their main solutions currently include a series of sensor networks, towers, drones, and powerful software that ties it all together - whose potential uses include protecting our troops on base, defending our energy infrastructure, combating wildfires, stopping human traffickers, creating a “virtual border” (a rare bipartisan idea), and fighting drug cartels. Many of these potential uses can directly save lives.

In addition to Anduril, a number of large technology incumbents (such as Salesforce, Microsoft, and Amazon) and leading startups (Applied Intuition, SpaceX, Scale.AI, and others) continue to provide their technologies to enable our defense and intelligence services.

To try to understand this world better, and as part of investing further in Anduril, I interviewed a number of former high level individuals involved with our country’s defense and related policy. A number of people told me they found these interviews interesting, so I am sharing them here[2]. 

I had a number of takeaways from these conversations. One of my biggest takeaways was the degree to which national defense was truly a bipartisan issue. Democrats and Republicans both believe that technology can be used to protect our families, friends, and neighbors at home as well as members of our military and intelligence communities overseas. Both sides believe that transparency into what is happening around the world is critical, and technologies that increase that transparency are ones we should invest in. And both sides agree that the days of America’s military having the best technology may be coming to a close, and that “business as usual” won’t save us. “I think we need a better ecosystem to work within, Retired Major General Vincent Coglianese told me. “You need to have this nimble relationship with technology.” He, and everyone else I talked to, believe that Anduril is an important player in that shift.

A subset of conversations are excerpted here with permission of the people I interviewed.

  1. Eric Snelgrove, former Professional Staff Member with the House Armed Services Committee and Minority Staff Lead of the Intelligence, Emerging Threats and Capabilities (IETC) Subcommittee.

  2. Katie Wheelbarger, former Assistant Secretary of Defense for International Security Affairs.

  3. Retired Major General Vince Coglianese, the former Commander of Marine Corps Installations Command.

  4. Tina Chong, former Army Captain.

Read the interviews here.


[1] I am actively looking to invest in Climate Change related companies, with an emphasis on software and data analytics related tooling, or carbon capture technologies.

[2] These discussions have been edited down for clarity and conciseness.

Monday, March 1, 2021

Back To The Office

US should be roughly done with COVID in 2021

Almost exactly a year ago, I wrote a blog post warning the tech community about a new coronavirus in China[0] and the likelihood we will see multiple viral waves.

By end of March, 2021, the US is estimated to have enough vaccine to cover roughly half its adult population for COVID. By end of July, Pfizer and Moderna alone should have supplied the US government with 600 million doses - enough to cover roughly the entire US population including children. Assuming it takes a month or two to distribute the vaccines and another month between shots (and a month to make some mistakes, because, why not?), most of the USA should be covered by sometime in the fall[1]. The country should start to reopen sometime during the summer (probably red states first) with no strong logical reason not to reopen roughly everything by end of year [2]. This means by end of 2021 we should anticipate most teams back in the office barring something unexpected.

Image source 

By end of 2021, a big enough portion of the developed world will be vaccinated to allow previously remote offices to reopen and work to return to normal [2]. As companies think ahead on the future of work, a few different models emerge. In this post we discuss models of work, the future of corporate travel, and an aside on the Bay Area & tech.


Many people now want to go back into an office

In mid-2020 multiple company surveys shared with me suggested tech company employees were overall content to be working from home. No commute! Flex time with kids! Can go be close to family! Were all seen as positives. Employees were more productive than ever at some companies based on rate of code commits or other (messy) metrics. However, COVID has also provided a distorted lens on this - people were scared to go see other people and all distractions were eliminated - no more gym time, hobbies, lunches or dinners with friends. The only thing people had was work or family.

CEOs of many of those same companies are now telling me that the majority of these same employees want to go back to the office. Employees want to see and build bonds and friendships with their coworkers, have a common culture, and work together. Some people also just want to get away from their families :) Not every employee wants to go back every day of the week, but many would welcome an office environment again.

Speed of back to work

A number of hardware and biotech companies have stayed open (or had subsets stay in office) alongside many other type of industrial, manufacturing (Tesla et al) , and agricultural companies. Recently some software companies have also reopened at 20% to 50% capacity or more - typically with mask, COVID testing, and distancing in office.

Through the course of 2021 office reopenings should accelerate and are likely to occur “slowly, and then all at once”. The most likely cause of office opening slow downs post-summer (when the US should be able to vaccinate anyone who wants a vaccine) will be unvaccinated employees who refuse the vaccine. 

The “we do not know enough” about the long term side effects of the vaccine will be a false argument - hundreds of millions of people will have had a vaccine by the fall and side effects to date in both practice (50% of Israel vaccinated, and >10% of the USA) and in clinical trials have been minimal and transitory - but that argument will be made. In Israel, a “green passport” allows only vaccinated people to go the gym or movie theaters and legislative proposals are moving forward to move the societal cost of non-vaccination to people in specific roles who refuse vaccination. The US may need to eventually either (1) remove liability to re-opening employers for employees who refuse what to date is a safe set of vaccines or (2) allow employers to suspend employees who refuse the vaccine depending on the type of work being done.  For example, different states have non-COVID immunization requirements for teachers in order to allow them into a classroom. [Update: see article on this here].

One can anticipate a number of future articles in the press about the lone, hold out unvaccinated employee going to work double-masked and trying to push the company to social distance (“we just don’t have enough data yet!” will be the false argument), while everyone around them is vaccinated and wants to move on with life. School and healthcare settings will likely have more politics and complexity around reopening.

Employers may at some point want to simply point out to unvaccinated workers they are at risk & can be vaccinated but need to be in the office and then roughly ignore them (you should talk to your legal team first of course). Frequent testing for unvaccinated employees (but not vaccinated ones) may be another option to allow the office to re-open to pre-COVID norms for those vaccinated. One Israeli legislative push is for unvaccinated people to have to pay for their own frequent testing in order to protect the rest of society. 

Models for Work

There are 3 future models for work, with lots of gray area in between. Each company will chose their own model from below. I anticipate most companies to choose models 1 or 2.

1. Full snap back to pre-COVID in-office work

In this approach most people are back in the office 5 days a week, with a subset of people taking a work from home day and an even smaller subset working fully remote ongoing. This is the same as the pre-COVID era, but probably with a larger proportion of the work force remote. Smaller companies (e.g. a few hundred people or less) on average are likely to adopt this approach. Larger companies (>1000) will probably already have a multi-office work force and some remote workers prior to COVID so are likely to have more flexible work styles (see below).

In general, I think a large number of early or mid-stage companies will "fully snap back" to in-office work (despite all the hype to the contrary).

2. Hybrid 

Hybrid models have 2 main approaches:

2A. Hybrid: 3 days in office for everyone, 2 flex days.

A number of companies I know are going back to work for 2-3 days a week, with flex (be in office or not) the other days for most teams. The key is that all the employees will be in the office on the same set of days - to maximize in person interactions and collaboration time. Google, Salesforce, and others appear to be moving in this direction.

One implication of this is most workers still need to live close to a major hub or office, but one can imagine longer commute times (e.g. live in wine country but commute to SF 2-3 times a week). So for San Francisco as an example, more people will migrate out of SF to wine country or Sacramento and then drive in a few times a week.

It is possible this style of working will stick (particularly for very large public companies where honestly the average employee probably isn't that hard charging) while some smaller teams may stick with it or decide to flip back to mainly in-office work.

2B. Hybrid: Mainly back in office, but “remote” is considered its own office.

An alternate (and potentially overlapping) model to this is to still have a set of core locations were much of the employee base goes to work, but “remote” is also considered a full fledged “office”. This is actually how older school tech companies like HP and Cisco had some of their teams prior to COVID. Facebook and others seem to be moving in this direction. The extreme version of this has the CEO and executive team fully distributed across these offices, or as part of the “remote office” versus semi-centralized in one of the offices.

One version I have heard talked about but not yet implemented is where a company is remote-first, but also has either rented out or built “coworking space” in major cities or clusters. No one has a permanent desk, but rather each day need to come in and set up a new spot. Given the heavy ergonomics (desk, monitor, mouse placement etc) I have seen people set up in a typical office environment, I think this version eventually collapses in the limit to simply having office space with pre-set spots. Lets see.

3. Remote only

In this version there are no physical offices operated by the company. People can either work from home or from a 3rd party co-working space. There is no longer any form of headquarters and the team can be truly distributed (although early on specific timezones are likely to be emphasized or selected for [4]).

One benefit of this approach are you can hire anyone around the world (although timezone clusters tend to form early) [4]. You also may end up paying less, on average, as you match local pay more closely than Bay Area pay (which largely siphons off income to landlords).

Negatives include the need to scale processes early in the life of the company (such as goal setting or OKRs, tight onboarding, a move to asynchronous communication, and lots of documentation). 30 person remote-first teams may have the processes of 300 person non-remote companies - so going remote first comes with some unexpected overhead to be successful. New projects are much harder to kick off truly remote, while at scale “turn the crank” projects seem to survive remote work more. There are also some unexpected challenges - e.g. if a remote-only company buys a “normal” company with offices - do they keep the offices? Are the employees of the acquired companies going to adapt to a remote-only culture? It is possible the pandemic has made this an easier transition than in the past.

Gitlab at this point is the canonical remote-first company pre-pandemic. It is notable that only 3 tech companies seemed to have ever hit the scale of many hundred employees or more as truly a remote first company pre-pandemic - Gitlab, Automaticc, and Zapier (most other examples were subscale, or not truly remote first - however there may be others?). It will be interesting to see if a generation of remote-only companies will now grow to scale after the pandemic forced them to adopt remote-only early in life.

Choosing a model

When choosing the model that is right for you - you should take into account what is best for your company (versus what is most popular with employees in a survey before COVID is over - this might not be the same thing). It is perfectly OK to chose 5 days a week back in the office.

For many early or mid stage companies, being in the office most or all of the time may be the best solution given the need to form culture, foment creativity, and work tightly in a collaborative manner.

For a later stage company the best model may be team or function dependent - teams working on creative new products may benefit from being in office together more, while those working on existing products may need less time in person together. Similarly, some functions will benefit more from hallway conversations and face to face time than others.

Irrespective of what model you chose above, what the CEO and executive team does will be emulated by a large portion of the company.

Trade-offs in remote versus in-office work

Process matters more remote

When you start scaling a company (e.g. growing from tens to eventual thousands of employees) you realize that lightweight processes can help an organization from breaking and parts of the company going off the rails. On average, remote-first or remote-only companies need to implement scaling processes at an order of magnitude smaller company scale than an in-office company. While a 30 or 100 person company may all cohabitat the same office space, a 500 person company undoubtedly has multiple offices in different and therefore needs to build tooling to allow for people to collaborate across offices and timezones. A remote-only company needs to start to implement the same process much earlier. Process may include things like OKRs/goal setting, heavier asynchronous communication and documentation, special multi-day in person “onsites” to allow for collaboration on new creative projects, and very crisp onboarding, cultural integration, and other areas. This means that being remote-only or remote-first comes with inevitable process costs (and in some cases benefits - e.g. all companies should have early goal setting and clear individual and team ownership of areas).

New projects versus scaling existing

Many CEOs I know feel that remote work is easier for existing, scaling projects or products that are in “turn the crank” mode, while new creative endeavors are easier with in person, rapid fire collaboration. This means it is easier to be a large company or specific subteams (with existing process and products) that is remote first, than a small company (which is still figuring it all out).

Synchronous vs asynchronous

In general remote (or distributed) teams have dramatically more asynchronous collaboration and communication than in person teams, for obvious reasons. This usually means more documentation, and use of tools that allow for non-real time communication.

Salary and compensation 

The majority of companies will differentially compensate employees based on their locations (for example Facebook, Twitter, Stripe, Gitlab, VMWare, and others all adjust pay - Reddit may be the rare holdout). A number of mid-market and large tech companies seem willing and eager to have employees move out of SF in particular, given its poor regulatory/government environment and high cost of living. I would not be surprised if more companies paid a “moving bonus” for people to move out of San Francisco or maybe the broader Bay Area. 

In parallel, more companies are hiring people not only out of the Bay Area, but also out of the USA - with hires in places like Argentina, Brazil, India, Poland, Romania, Serbia, Ukraine and other countries were cost of living and base salaries tend to be significantly lower.

Differential compensation tends to reflect:

1. It is better for companies to have people near or at hubs for communication and collaboration - so companies are willing to pay a premium if an employee continues to live near HQ or a major hub (typically Bay Area, New York, LA).

2. Local labor markets & costs of living differ. Many companies are paying a “moving” bonus and then cutting ongoing pay by anywhere from 10-30% depending on the new employee location, the local labor market, and cost of living.

3. This is already common practice and is non-controversial. Market-base salary adjustment are a long standing practice in every industry including tech. The “I want to have my (lifestyle) cake and eat it (Bay Area salary) too” view is likely misplaced relative to historical industry norms.

San Francisco as a special case

In October 2020, roughly half of US unicorns were in the Bay Area, but literally all 5 of the private unicorn companies announcing new remote first plans were based there (with 4 of 5 in San Francisco). This seems to be driven by a mix of high cost of living (lack of affordable housing), ever-increasing petty crime, homelessness, drug abuse issues, terrible public school policies, and general poor-governance. In parallel, some employees based in the Bay Area strike many company leaders as not only more expensive but also more entitled and politically-motivated (versus company-mission motivated) at work than their non-Bay Area counterparts. Many CEOs also view San Francisco as a city that will continue to raise taxes, for example the taxes on gross receipts and other add-ons are costing multiple companies headquartered in San Francisco thousands or tens of thousands per employee versus being located elsewhere - including employees who do not even live in San Francisco. 

This mix of expense and bad governance will continue to drive companies out of San Francisco (although many may relocate within the Bay Area itself - for example South San Francisco is much more business friendly). Some may go remote first, and some may simply leave. This will, of course, take many years to play out. The epicenter of tech in the Bay Area has moved over time from San Jose/Santa Clara (1970s-1990s) to Palo Alto/Mountain View (2000s) to San Francisco (2010-now). It will be interesting to watch if it sticks, or where it goes next in the Bay Area.


[0] It is nice to be able to now write about re-opening, even if it will take some time.

[1] Of course, some people will decline the vaccine and a small number may be too remote to reach. Anticipate much hand-wringing about how we “must protect the unvaccinated” who *chose* not get vaccinated. An approach likes Israeli’s might work- a “green passport” that only allows people who are vaccinated to go to gyms, movies etc. as a positive incentive to get vaccinated.

[2] We will eventually (probably sometime during end of 2021) need societally to ignore the public health voices, special interest or political groups that will claim it is never safe enough to truly reopen because “variants”[3] or “only 90% of at-risk people are protected” or “ethics & equity” (really politics) or some other reason. See for example the lack of school reopenings in San Francisco. It would be a large net positive for society if we avoid TSA-like theatre for most aspects of life in the years to come.

[3] At this point all the vaccines have protected strongly from death for every vaccine and every variant (including the “South African” one). It is useful to remind oneself that the primary goal is to prevent deaths and hospitalizations. Unfortunately, as a worst case to date, while the AstraZeneca vaccine protected against deaths and severe disease in recent South African trials, its efficacy against mild to moderate COVID was weakened. The AZ vaccine has generally had lower neutralizing antibody titers than some of the other vaccines, so we will see data from others in the coming weeks. Some vaccine companies have already started trials with boosters to address variants in case needed. The most likely scenario is society will not be impacted much be the existing set of variants, however that is the largest near-term real risk to reopening.

[4] Time zones make a big difference in the ability to coordinate the synchronous part of work. For sales this does not matter as much, but for product development it often does. So often small to mid-size companies tend to collapse product development into a smaller set of time zones after trying to have people everywhere. Once a company gets big enough that autonomous teams can split off onto specific projects, multi-timezone efforts are easier.


You can order the High Growth Handbook here. Or read it online for free.



Hiring Employees and Executives



Startup life:

Raising Money