Finding the Goldilocks zone for applied AI

While Elon Musk and Mark Zuckerberg debate the dangers of artificial general intelligence, startups applying AI to more narrowly defined problems such as accelerating the performance of sales teams and improving the operating efficiency of manufacturing lines are building billion-dollar businesses. Narrowly defining a problem, however, is only the first step to finding valuable business applications of AI.

To find the right opportunity around which to build an AI business, startups must apply the “Goldilocks principle” in several different dimensions to find the sweet spot that is “just right” to begin — not too far in one dimension, not too far in another. Here are some ways for aspiring startup founders to thread the needle with their AI strategy, based on what we’ve learned from working with thousands of AI startups.

 “Just right” prediction time horizons

Unlike pre-intelligence software, AI responds to the environment in which they operate; algorithms take in data and return an answer or prediction. Depending on the application, that prediction may describe an outcome in the near term, such as tomorrow’s weather, or an outcome many years in the future, such as whether a patient will develop cancer in 20 years. The time horizon of the algorithm’s prediction is critical to its usefulness and to whether it offers an opportunity to build defensibility.

Algorithms making predictions with long time horizons are difficult to evaluate and improve. For example, an algorithm may use the schedule of a contractor’s previous projects to predict that a particular construction project will fall six months behind schedule and go over budget by 20 percent. Until this new project is completed, the algorithm designer and end user can only tell whether the prediction is directionally correct — that is, whether the project is falling behind or costs are higher.

Even when the final project numbers end up very close to the predicted numbers, it will be difficult to complete the feedback loop and positively reinforce the algorithm. Many factors may influence complex systems like a construction project, making it difficult to A/B test the prediction to tease out the input variables from unknown confounding factors. The more complex the system, the longer it may take the algorithm to complete a reinforcement cycle, and the more difficult it becomes to precisely train the algorithm.

While many enterprise customers are open to piloting AI solutions, startups must be able to validate the algorithm’s performance in order to complete the sale. The most convincing way to validate an algorithm is by using the customer’s real-time data, but this approach may be difficult to achieve during a pilot. If the startup does get access to the customer’s data, the prediction time horizon should be short enough that the algorithm can be validated during the pilot period.

For most of AI history, slow computational speeds have severely limited the scope of applied AI.

Historic data, if it’s available, can serve as a stopgap to train an algorithm and temporarily validate it via backtesting. Training an algorithm making long time horizon predictions on historic data is risky because processes and environments are more likely to have changed the further back you dig into historic records, making historic data sets less descriptive of present-day conditions.

In other cases, while the historic data describing outcomes exists for you to train an algorithm, it may not capture the input variable under consideration. In the construction example, that could mean that you found out that sites using blue safety hats are more likely to complete projects on time, but since that hat color wasn’t previously helpful in managing projects, that information wasn’t recorded in the archival records. This data must be captured from scratch, which further delays your time to market.

Instead of making singular “hero” predictions with long time horizons, AI startups should build multiple algorithms making smaller, simpler predictions with short time horizons. Decomposing an environment into simpler subsystems or processes limits the number of inputs, making them easier to control for confounding factors. The BIM 360 Project IQ Team at Autodesk takes this small prediction approach to areas that contribute to construction project delays. Their models predict safety and score vendor and subcontractor quality/reliability, all of which can be measured while a project is ongoing.

Shorter time horizons make it easier for the algorithm engineer to monitor its change in performance and take action to quickly improve it, instead of being limited to backtesting on historic data. The shorter the time horizon, the shorter the algorithm’s feedback loop will be. As each cycle through the feedback incrementally compounds the algorithm’s performance, shorter feedback loops are better for building defensibility. 

“Just right” actionability window

Most algorithms model dynamic systems and return a prediction for a human to act on. Depending on how quickly the system is changing, the algorithm’s output may not remain valid for very long: the prediction may “decay” before the user can take action. In order to be useful to the end user, the algorithm must be designed to accommodate the limitations of computing and human speed. 

In a typical AI-human workflow, the human feeds input data into the algorithm, the algorithm runs calculations on that input data and returns an output that predicts a certain outcome or recommends a course of action; the human interprets that information to decide on a course of action, then takes action. The time it takes the algorithm to compute an answer and the time it takes for a human to act on the output are the two largest bottlenecks in this workflow. 

For most of AI history, slow computational speeds have severely limited the scope of applied AI. An algorithm’s prediction depends on the input data, and the input data represents a snapshot in time at the moment it was recorded. If the environment described by the data changes faster than the algorithm can compute the input data, by the time the algorithm completes its computations and returns a prediction, the prediction will only describe a moment in the past and will not be actionable. For example, the algorithm behind the music app Shazam may have needed several hours to identify a song after first “hearing” it using the computational power of a Windows 95 computer. 

The rise of cloud computing and the development of hardware specially optimized for AI computations has dramatically broadened the scope of areas where applied AI is actionable and affordable. While macro tech advancements can greatly advance applied AI, the algorithm is not totally held hostage to current limits of computation; reinforcement through training also can improve the algorithm’s response time. The more of the same example an algorithm encounters, the more quickly it can skip computations to arrive at a prediction. Thanks to advances in computation and reinforcement, today Shazam takes less than 15 seconds to identify a song. 

Automating the decision and action also could help users make use of predictions that decay too quickly to wait for humans to respond. Opsani is one such company using AI to make decisions that are too numerous and fast-moving for humans to make effectively. Unlike human DevOps, who can only move so fast to optimize performance based on recommendations from an algorithm, Opsani applies AI to both identify and automatically improve operations of applications and cloud infrastructure so its customers can enjoy dramatically better performance.

Not all applications of AI can be completely automated, however, if the perceived risk is too high for end users to accept, or if regulations mandate that humans must approve the decision. 

“Just right” performance minimums

Just like software startups launch when they have built a minimum viable product (MVP) in order to collect actionable feedback from initial customers, AI startups should launch when they reach the minimum algorithmic performance (MAP) required by early adopters, so that the algorithm can be trained on more diverse and fresh data sets and avoid becoming overfit to a training set.

Most applications don’t require 100 percent accuracy to be valuable. For example, a fraud detection algorithm may only immediately catch five percent of fraud cases within 24 hours of when they occur, but human fraud investigators catch 15 percent of fraud cases after a month of analysis. In this case, the MAP is zero, because the fraud detection algorithm could serve as a first filter in order to reduce the number of cases the human investigators must process. The startup can go to market immediately in order to secure access to the large volume of fraud data used for training their algorithm. Over time, the algorithms’ accuracy will improve and reduce the burden on human investigators, freeing them to focus on the most complex cases.

Startups building algorithms for zero or low MAP applications will be able to launch quickly, but may be continuously looking over their shoulder for copycats, if these copycats appear before the algorithm has reached a high level of performance. 

There’s no one-size-fits-all approach to moving an algorithm from the research lab to the market.

Startups attacking low MAP problems also should watch out for problems that can be solved with near 100 percent accuracy with a very small training set, where the problem being modeled is relatively simple, with few dimensions to track and few possible variations in outcome.

AI-powered contract processing is a good example of an application where the algorithm’s performance plateaus quickly. There are thousands of contract types, but most of them share key fields: the parties involved, the items of value being exchanged, time frame, etc. Specific document types like mortgage applications or rental agreements are highly standardized in order to comply with regulation. Across multiple startups, we have seen algorithms that automatically process these documents needing only a few hundred examples to train to an acceptable degree of accuracy before additional examples do little to improve the algorithm, making it easy for new entrants to match incumbents and earlier entrants in performance.

AIs built for applications where human labor is inexpensive and able to easily achieve high accuracy may need to reach a higher MAP before they can find an early adopter. Tasks requiring fine motor skills, for example, have yet to be taken over by robots because human performance sets a very high MAP to overcome. When picking up an object, the AIs powering the robotic hand must gauge an object’s stiffness and weight with a high degree of accuracy, otherwise the hand will damage the object being handled. Humans can very accurately gauge these dimensions with almost no training. Startups attacking high MAP problems must invest more time and capital into acquiring enough data to reach MAP and launch. 

Threading the needle

Narrow AI can demonstrate impressive gains in a wide range of applications — in the research lab. Building a business around a narrow AI application, on the other hand, requires a new playbook. This process is heavily dependent on the specific use case on all dimensions, and the performance of the algorithm is merely one starting point. There’s no one-size-fits-all approach to moving an algorithm from the research lab to the market, but we hope these ideas will provide a useful blueprint for you to begin.

Inside Nickelodeon’s Teenage Mutant Ninja Turtles VR Interview Experience

Last month at San Diego Comic Con, I fulfilled my childhood (and let’s be honest, current) dream of stepping inside a NickToon. In Nickelodeon Entertainment Lab’s Rise of the Teenage Mutant Ninja Turtles VR Interview Experience, I interacted with turtles Mikey and Donnie—voiced live by series talent, Brandon Mychal Smith and Josh Brener, respectively.

I stood against a green screen, selected an avatar (I chose Arnold from Hey Arnold!), put on an Oculus Rift headset, and was transported onto a New York City rooftop. I inhabited a 3D form, but as I looked ahead, I saw two 2D turtles. I interviewed the characters in real time, and their movements perfectly corresponded to their answers–there was no discernable lag.

Given that AI isn’t able to match the conversational speed and nimbleness of real humans just yet, companies like Oculus are experimenting with using live actors in their VR experiences to approximate in-person interactions. However, this was the first time I’ve experienced a live melding of 3D and 2D worlds in VR.

Behind the scenes, this “magic” was made possible by four computers, two puppeteers, two voice actors and a team of eight people running the various stations. (The crew were all wearing ninja bandanas, naturally). Adobe Character Animator, Epic Unreal Engine, and NewTeK NDI were all used to create the interview experience. The pièce de résistance was the bright purple keyboard that was piloted by two team members (one for each turtle). Each key featured a different animated pose, and so the actions of the turtles could be “animated” in real time, akin to playing a symbol piano.

According to Chris Young, SVP of the Entertainment Lab, the impetus of the activation was that his team was looking for an innovative way to help socialize Nickelodeon’s Rise of the Teenage Mutant Ninja Turtles at SDCC and so they came up with the idea of a VR press junket.

Mikey and Donnie in Nickelodeon’s “Rise of the Teenage Mutant Ninja Turtles Live” virtual reality interview experience

The Entertainment Lab has been exploring all aspects of virtual cinema pipelines with animated characters; using full body and facial performance capture, and doing real-time playback in game engines. Per Chris: “Whether it’s streaming live performances into virtual experiences or recorded for a more traditional linear output, these techniques create another tool that we can use to create compelling content for our audience.”   

Even without the bandanas, the team had great synergy: “The best part about my team is that everyone brings a different expertise. From artists to engineers, the combination of skills and background made this creatively and technically possible” said Chris.

Although I teased the turtles for being “flat” in my interview, I love that this creative choice was made. It would have been unsettling to suddenly see what is 2D in 3D.

For example, I enjoy the premise and plot of Virtual Rick-Ality, but one of my main criticisms of the game is the connection between assets. Rick and Morty the animated series is 2D, but the game makes them into bulbous characters. This disconnect is off-putting and hurts the immersion of the game.

Although the VR Interview Experience was created specifically for Comic-Con and Nickelodeon doesn’t currently have plans to release a version to the public, they are toying around with other activations for fans.

However, VR isn’t where exploration ends for Nickelodeon, they’re also dabbling in AR. Nick’s new SCREENS UP initiative, first launched at the Nickelodeon Kids Choice Sports Awards, allows viewers to hold their mobile devices (or more likely, parent’s mobile devices) up to the screen while watching to reveal hidden AR content.

This app-driven TV and mobile experience is one of the first in the U.S. designed for kids and families. Expect to see more hybrids like this in the future, as it encourages the watching of live TV as well as app downloads; two desirable outcomes for television content providers.

Overall, the best part of my Ninja Turtles Comic Con demo was how silly and good-natured it all was. With heavy headsets and an onslaught of first person shooters, VR can be kind of serious! This felt social and delightful — the improvisational nature of the conversation allowed me to forget I was in a digital space, I was just hanging out with some dope turtles. It’s a credit to Nickelodeon that they could make something this complex look like child’s play.

The “Rise of the Teenage Mutant Ninja Turtles Live” virtual reality experience was built entirely in-house by Nickelodeon’s Entertainment Lab, which spearheads long-range research and development efforts around new technologies for Nickelodeon and its audience. Chris Young is the SVP of the Entertainment Lab, overseeing its creation. The new Rise of the Teenage Mutant Ninja Turtles series premieres September 17th on Nickelodeon.

California may mandate a woman in the boardroom, but businesses are fighting it

California is moving toward becoming the first state to require companies to have women on their boards –assuming the idea could survive a likely court challenge.

Sparked by debates around fair pay, sexual harassment and workplace culture, two female state senators are spearheading a bill to promote greater gender representation in corporate decision-making. Of the 445 publicly traded companies in California, a quarter of them lack a single woman in their boardrooms.

SB 826, which won Senate approval with only Democratic votes and has until the end of August to clear the Assembly, would require publicly held companies headquartered in California to have at least one woman on their boards of directors by end of next year. By 2021, companies with boards of five directors must have at least two women, and companies with six-member boards must have at least three women. Firms failing to comply would face a fine.

“Gender diversity brings a variety of perspectives to the table that can help foster new and innovative ideas,” said Democratic Sen. Hannah-Beth Jackson of Santa Barbara, who is sponsoring the bill with Senate President Pro Tem Toni Atkins of San Diego.”It’s not only the right thing to do, it’s good for a company’s bottom line.”

Yet critics of the bill say it violates the federal and state constitutions. Business associations say the rule would require companies to discriminate against men wanting to serve on boards, as well as conflict with corporate law that says the internal affairs of a corporation should be governed by the state law in which it is incorporated. This bill would apply to companies headquartered in California.

Jennifer Barrera, senior vice president of policy at the California Chamber of Commerce, argued against the bill and said it only focuses “on one aspect of diversity” by singling out gender.

“This bill basically mandates that we hire the woman above anybody else who we may be fulfilling for purposes of diversity,” she said at a hearing.

Similarly, a legislative analysis of the bill cautioned that it could get challenged on equal protection grounds, and that it would be difficult to defend, requiring the state to prove a compelling government interest in such a quota system for a private corporation.

Five years ago, California was the first state to pass a resolution, authored by Jackson, calling on public companies to increase gender diversity. In response, about 20 percent of the companies headquartered in the state followed through with putting women on their boards, according to the research firm Board Governance Research. But the resolution was non-binding and expired in December 2016.

Other countries have been more proactive. Norway in 2007 was the first country to pass a law requiring 40 percent of corporate board seats be held by women, and Germany set a 30 percent requirement in 2015. Spain, France and Italy have also set quotas for public firms.

In California, smaller companies have fewer female directors. Out of 50 companies with the lowest revenues, 48 percent have no female directors, according to Board Governance Research. Only 8 percent of their board seats are held by women.

The 2017 study said larger companies did a better job of appointing women, with all 50 of the highest-revenue companies having at least one female director and 23 percent of board seats held by women.

“The main issue is still that a lot of companies headquartered here don’t have women on their boards,” said Annalisa Barrett, clinical professor of finance at the University of San Diego’s School of Business. “We quite often like to think of California as progressive and a leader on social issues, so that’s kind of disappointing.”

Barrett publishes an annual report of women on boards in California. Public companies are major employers in the state, and their financial performance has a big impact on public pension funds, mutual funds and investment portfolios. “Financial performance does really impact the broader community,” she said.

The National Association of Women Business Owners, sponsor of the bill, says an economy as big as California’s ought to “set an example globally for enlightened business practice.” In a letter of support, the association cites studies that suggest corporations with female directors perform better than those with no women on their boards.

One University of California, Davis study did find that companies with more women serving on their boards saw a higher return on assets and equity, but the author acknowledges this may not suggest a cause-and-effect.

Blind loyalty

There is a secret behind every open office in Silicon Valley — and it isn’t the drain on productivity.

Tech companies have been the vanguards for pushing corporate culture forward toward “radical transparency.” Mark Zuckerberg works in a fully transparent four-walled glass office surrounded by the rest of Facebook. Valve got rid of managers and titles so everyone can be their own boss. Startup founders host weekly town halls, Friday all-hands, and AMAs. Companies go to painstaking lengths to signal that they trust their employees – to show that this is your company.

But while your company might adopt an open floor plan and give out free snacks so you can feel closer to your coworkers, they likely don’t want you knowing how much they make, who is affected by the impending layoffs, or whether executives are making the right decisions.

The open office has never been more closed, and tech companies are no different than old corporate America in their authoritarian approach to controlling how their employees should think about issues that matter in the workplace. In fact, it may even be more insidious because it’s tucked away behind the veneer of a cheerful, open office.

This is what makes social network Blind so fascinating. Raw and unfiltered, Blind is the antithesis to HR’s utopic vision of a manageable and orderly corporate culture. Instead, it operates outside the walled gardens of IT with no rules and no official corporate supervision.

With Blind, users are completely anonymous, but are required to submit a verified work email to join a company channel. Inside, they are able to freely ask, discuss, prod, and complain without fear of retribution or judgment.

In short, it’s HR’s worst nightmare, and it’s wildly successful.

Building a compelling social product

Blind’s engagement numbers are staggering. It has over 2 million users, including 43K at Microsoft, 28K at Amazon, and 10K at Google. In South Korea, half of all employees at companies over 200 people are active monthly. The typical monthly active user logs in three to four times per day and spends 35 minutes using the app. At the height of the Susan Fowler scandal, Uber employees were spending almost 3 hours a day on Blind. All that, and the entire company is 38 people.

At the heart of Blind’s magic is something universal to every person who has ever been employed — the duality between our personal selves and our “work” selves, and the human drive to be both intimate and in control of our relationships. There is no place more difficult to navigate this duality than the workplace, where we want to feel loved and understood, but also respected.

Hierarchy, politics, and negative career impacts burden conversations about difficult topics, and so Blind tears these barriers down one employee at a time, affording a space for uninhibited dialogue. More importantly, Blind succeeds as a resource for questions not only company-related, but also around career, family, and life decisions.

Blind is in many ways an evolution of a long lineage of ideas in social networking. It’s unique achievement is the recombination of these different ideas to create a platform that is both a safe space for free and open conversation (via anonymity), along with a vetted, contextually relevant community (via workplace email authentication).

Let’s walk though each of these categories to understand Blind’s success.

Lack of Context (Anonymous + Individual/Personal) – Companies like Yik Yak, Secret, and Whisper pioneered the anonymous social network on the consumer side. However, they were beleaguered by cyberbullying, and served more as a digital exhaust pipe for teenage angst and trolling. Perhaps the most successful semi-anonymous social network today is Reddit, where legions of loyal community members cover every topic imaginable. However, what all of these anonymous communities lack is the critical element of shared context and circumstance.

Put another way, your fellow community members on Reddit may share your interest in ice fishing, but they likely will not understand who you are. As Blind cofounder Kyum Kim puts it, “it’s hard for someone to complain on Reddit about feeling poor while making $200K a year without fear of backlash, but on Blind, your coworkers are in the same income bracket, and likely similar education levels, neighborhoods, etc. They can empathize with your situation.” On Blind, there is a single community (your workplace) that spans multiple topics, and there’s a baseline, tacit understanding of each other’s life circumstances, allowing for deeper conversations.

Self-Promoting (Non-Anonymous + Individual/Personal) – LinkedIn and Quora are useful professional platforms, but because individuals and brands are the stars of these platforms, posturing and self-promotion can be quite frequent. When you ask a question on Quora, you are submitting your inquiry to a body of self-proclaimed experts. While many responses can be genuine, the ultimate currency that drives the platform is credibility and brand building, which inhibit authentic and vulnerable conversations from occurring.

Self-Censored (Non-Anonymous + Employee/Work) – On the enterprise side, Yammer, Jive, and recently Slack have attempted to upgrade the creaky company intranet into the enterprise social network. While these tools might make it easier to connect to your coworkers, the conversations happening on these platforms are no different than before – ultimately, these tools are designed to get work done, not for questioning, debating, or reflecting on how work should be. Conversations about sensitive subjects (e.g. how to deal with a bad manager) are unlikely to happen on a non-anonymous, corporate-sanctioned platform where that same bad manager might well be watching.

Finally, we have Blind. The platform strikes a balance between the freedom of anonymity and the context of a shared workplace. The result is a forum for surprisingly rich, relevant, and authentic conversations. While company channels are accessible only to insiders, a look at Blind’s public site (where you still need a verified work email, but you can chat with anyone outside your company) reveals a flavor for the types of conversations that are possible. An engineer at Amazon recently posted about how to deal with a mid-life crisis, with 42 responses of encouragement and advice. Another employee moving from India has a wife suffering from depression and is seeking help navigating the US healthcare system.

It turns out that where we work is a good proxy for who we are, and our coworkers have been an untapped community of wisdom.

Trust and safety

Catalin205 via Getty Images

Blind is by no means perfect. Like all online platforms and particularly anonymous ones, it invites its share of trolls. One look at the “Relationships” section on Blind’s public site and you’ll find questions about how to deal with one-night stands with coworkers and a poll asking guys how many girls they’ve slept with before marriage. While these questions could certainly have come from a genuine place, they are easy fodder for trolls, and the ensuing conversations can be alienating and provide an unnecessary megaphone for toxic bro culture.

Blind acknowledges that these issues exist, but claim that they happen less frequently inside company channels. Because users authenticate with their work emails, cofounders Sunguk and Kim believe that Blind users feel a greater sense of responsibility to each other because they are engaging a real community with shared context and goals.

The vast terrain of cyberspace might suffer from the tragedy of the commons and moral hazard, but within your workplace channel on Blind, your digital community maps onto a physical community – even though you are anonymous. This is evidenced by the successful self-policing on the platform, where 0.5% of all posts have been removed (higher than average for a social media platform), and all of these originated from user-generated flags.

A More Perfect Union

Blind’s success illuminates a reality that is often overlooked: corporations aren’t naturally democratic or transparent. While there are platforms to discuss our roles as individual working professionals (e.g. LinkedIn), there are very few places to gather and organize as employees of companies to collectively bargain for a better workplace.

This is by design. HR, the supposed watchdog of employee wellness, is neither elected nor truly representative, as they must balance the competing goals of being a third party resource for employees while also protecting the company against its employees.

Companies will always be incentivized to maintain an asymmetry of information. Friday all-hands and town halls are heavily scripted by companies. Rarely do we see anyone describing a healthy, transparent culture as a place where employees are freely conversing amongst themselves.

For companies with something to hide, the idea of a public square where conversations happen freely should be alarming. Blind has already been at the center of exposing two major scandals (e.g. the “nut rage” incident by a Korean Air executive and the news that Lyft was spying on its users.)

Blind picks up where labor unions left off and where HR has failed — to serve as a safeguard against corporate overreach, and to provide a protected space for employees to collaborate around solutions to improve the workplace.

A truly open office

For companies, Blind’s rise shouldn’t be seen as bad news. Blind can be a rich source of insight where HR software falls short. While employee engagement surveys have become popular in HR circles (and a crop of well-funded HR tech companies have consequently flooded the market), these practices suffer from the same issues of hosting a town hall. The company decides on the questions asked and interprets the answers given. With Blind, for the first time, HR and executives will have a pulse on employee sentiment that is both real-time and authentic. As Moon puts it, “no company is perfect, and if it was, Blind would not need to exist.”

In short, Blind understands more about your employees than anything in your HR stack.

Where does Blind go from here? Moon and Kyum believe they’re just getting started. Today, Blind is only available in the U.S. and South Korea, and it has been focused on tech companies. Their push into more traditional industries is showing some early signs of success with Johnson & Johnson, Dow Chemical, Barclays, and the US Navy coming online recently. There is still work to do in cleaning up different communities to ensure that conversations are inclusive and not alienating. And of course, Blind has to find a path to becoming a sustainable, revenue-generating company without compromising its integrity with users.

But one can only imagine the potential for Blind if it continues on its path upwards — the anonymous social network that understands who you are, the pulse survey that is authentic and real-time, and the first truly safe and open office made for employees, by employees.

Supergiant VC rounds aren’t just raised in China

In the venture capital market, big is in. Firms are raising significant sums to finance a growing number of large startup funding rounds.

In July, there were 55 venture rounds, worldwide, which topped out at $100 million or more, totaling just over $15 billion raised in nine and 10-figure mega-rounds alone. This set a record for venture dealmaking.

We’ve already identified approximately when the uptick in huge VC rounds began: toward the tail end of 2013. But where in the world are all the companies raising these supergiant venture capital rounds?

In response to coverage of July’s record-breaking numbers, many commenters were quick to point out that startups based in China raised six of the top 10 largest rounds from last month.

Indeed, on a recent episode of the Equity podcast discussing the supergiant round phenomenon, Chinese startups’ position in the market was a hot topic of conversation. Someone suggested that a series of large venture rounds in China may have preceded the run-up in supergiant rounds being raised by U.S. startups.

At least in the realm of nine and 10-figure venture rounds, that doesn’t appear to be the case. The chart below breaks down the monthly count of supergiant rounds by the company’s country of origin.

Here is what this data suggests:

  • The first major run-up in nine-figure dealmaking took place in the U.S. around Q1 2014, whereas in China that first run-up didn’t occur until Q4 2014.
  • Especially in the last 24 months or so, supergiant round volume in China and the U.S. is highly correlated, perhaps implying competition in the market.
  • We can see, very clearly, the mini-crash in the U.S. through the second half of 2015. For its part though, China hasn’t yet had a serious “crash” in supergiant rounds during this cycle.
  • Startups outside the U.S. and China are beginning to raise supergiant rounds at a faster rate, although the uptick is significantly less dramatic.

What’s less obvious in the chart above is just how quickly China became a mega-round powerhouse. The chart below plots the same data as above, except this format shows what percent of mega-rounds originated in each market. Additionally, rather than displaying somewhat noisy monthly amounts, we aggregated data in six-month increments.

After the start of 2013, it only took a couple of years for Chinese companies to consistently account for roughly 30 to 40 percent of the $100 million-plus VC rounds raised in any given six-month period.

This also reinforces a trend shown in the prior chart: since the beginning of 2017, Chinese startups and U.S. startups are raising roughly the same number of supergiant venture rounds as one another. That number has risen fairly consistently over time.

Before concluding, it’s worth mentioning that our definition of “supergiant” is ultimately arbitrary. Indeed, $100 million is just a tidy, round-numbered threshold to measure against. Our findings would be similar (if somewhat less dramatic) if we counted, say, the set of rounds raising $50 million or more.

The important underlying trend is that round sizes are getting larger on average. And a supergiant wave of money ultimately lifts all rounds, at least a little bit.

Stay up to date with recent funding rounds, acquisitions and more with the Crunchbase Daily.

Anonymous deals with its QAnon branding problem

When you're a notorious hacking entity like Anonymous, and a pro-Trump conspiracy cult (QAnon) steals your branding (while claiming you're the impostor), the obvious thing to do is declare cyberwar. That's exactly what Anonymous did this past week in…

Offering a white-labeled lending service in emerging markets, Mines raises $13 million

Emerging markets credit startup has closed a $13 million Series A round led by The Rise Fund, the global impact fund formed by private equity giant TPG, and 10 others, including Velocity Capital.

Mines provides business to consumer (B2C) “credit-as-a-service” products to large firms.

“We’re a technology company that facilitates local institutions — banks, mobile operators, retailers — to offer credit to their customers,” Mines CEO and co-founder Ekechi Nwokah told TechCrunch.

Most of Mines’ partnerships entail white-label lending products offered on mobile phones, including non-smart USSD devices.

With offices in San Mateo and Lagos, Mines uses big-data (extracted primarily from mobile users) and proprietary risk algorithms “to enable lending decisions,” Nwokah explained.

“We combine a strong AI technology with full…deployment services — disbursement…collections, payments, loan management, and regulatory — wrap it up in a box, give it to our partners and then help them run it,” he said.

Mines’ typical client is a company “that has a large customer base and wants to avail credit to that customer base,” according to Nwokah. The startup generates revenue from fees and revenue share with partners.

Mines started operations in Nigeria and counts payment processor Interswitch and mobile operator Airtel as current partners. In addition to talent acquisition, the startup plans to use the Series A to expand its credit-as-a-service products into new markets in South America and Southeast Asia “in the next few months,” according to its CEO.

Mines sees itself as a “hardcore technology company based in Silicon Valley with a global view,” according to Nwokah. “At the same time, we’re very African,” he said.

The startup’s leadership team is led by three Nigerians — Nwokah, Chief Scientist Kunle Olukotun and MD Adia Sowho. The company came together after Olukotun (then and still a Stanford professor) and Nwokah (a then-AWS big data specialist) met in Palo Alto in 2014.

Looking through the lens of their home country Nigeria, the two identified two problems in emerging markets: low access to credit across large swaths of the population and insufficient tools for big institutions to put together viable consumer lending programs.

Due to a number of structural factors in these markets, such as low regulatory support, lack of credit data and tech support, “there’s no incentive for many banks and institutions to take risk on a retail lending business,” according to Nwokah.

Nwokah sees Mines’ end user market as “the more than 3 billion adults globally without access to credit,” and its direct client market as big “banks, retailers and mobile operators…who want to power digital credit tailored to these markets.”

Mines views itself as different from the U.S.’s controversial payday lenders by serving different consumer needs. “If you live in a country where your salary is not guaranteed every month, where you don’t have a credit card…where you have to pay upfront cash for almost everything you do, you need cash,” he said

The most common loan profile for one of Mines’ partners is $30 at 15 percent flat for a couple of weeks.

Nwokah wouldn’t name specific countries for the startup’s pending South America and Southeast Asia expansion, but believes “this technology is scalable across geographies.”

As part of the Series A, Yemi Lalude from TPG Growth (founder of The Rise Fund) will join Mines’ board of directors.

On a call with TechCrunch, Lalude named the company’s ability to “drive financial inclusion within a matter of seconds from mobiles devices,” their “local execution on the ground” and model of “partnering with many large organizations with their own balance sheets” as reasons for the investment commitment.

With Mines’ pending Asia and South America move they join Nigerian tech companies and data analytics firm Terragon Group, who have expanded or stated plans to expand internationally this year.


As California burns, climate goals may go up in smoke — even after the flames are out

As crews across California battle more than a dozen wildfires — including the largest in state history — the blazes are spewing enough carbon into the air to undo some of the good done by the state’s climate policies.

What’s even worse: Climate-warming compounds that will be released by the charred forests long after the fires are extinguished may do more to warm up the planet than the immediate harm from smoky air.

Scientists say that only about 15 percent of a forest’s store of carbon is expelled during burns. The remainder is released slowly over the coming years and decades, as trees decay.That second hit of carbon, experts say, contains compounds that do more to accelerate climate change than those from the original fire. And future fires over previously burned ground could make climate prospects even more bleak.

“The worst possible situation is the fire that comes through and kills everything,” said Nic Enstice, regional science coordinator for the Sierra Nevada Conservancy. “Then, 10 or 15 years later, another fire comes through and releases all the carbon left in the trees on the ground. That’s really bad.”

It’s a scenario that could explode at any time. Enstice cited a research paper published this year that laid out a chilling tableau: California has more than a 120 million dead trees strewn around its mountain ranges, with the southern Sierra hardest hit.

When fires hit those downed trees, the state will begin to experience “mass fires” spewing plumes of carbon. The resulting conflagrations, according to the researcher, will be almost unimaginable.

“The emissions from those fires will be unlike anything we will have ever seen,” Enstice said. “And you won’t be able put it out.”

Computing the carbon released from the fires so far this year will not happen soon. The National Aeronautics and Space Administration flies planes through smoke plumes, gathering data, but air traffic over wildfires is tightly restricted. Scientific research is not a top priority when fires are threatening towns.

But some preliminary data is available now.

One method uses inventories of existing forests — surveying how many trees and which type. Those records are updated every 10 years. Researchers then overlay infrared images captured from satellites that show what’s burning and at what intensity. From that, predictions can be made about carbon emissions on any given day. Scientists say that emissions from burned forests are one of the most virulent types, called black carbon.

According to the most recent accounting from the state Air Resources Board, California’s annual black carbon discharge — excluding wildfires — are equal to emissions from about 8 million passenger vehicles driven for one year. Not a small number. But when the state calculates the same annual average of black carbon coming solely from wildfires, it’s the equivalent of nearly 19 million additional cars on the road.

With year-round fire seasons and fire intensity off the charts, state officials admit that wildfires could set back California’s myriad policies to offset the impacts of climate change. “It’s significant,” Enstice said. “We don’t have a lot of data to measure yet, we’re still using primitive tools. But everyone is gearing up to study this.”

This article is republished courtesy of CALMatters.