This Is The Future Of Artificial Intelligence
It’s widely known by now that the U.S. and global economy are being profoundly re-shaped by software technology. Human jobs are being eaten by software, specifically Artificial Intelligence (AI) algorithms able to ingest and analyze massive volumes of data to inform and remotely control better process management decisions, more efficient outcomes. Our political leaders don’t seem up to the policy challenges of job displacement — at least not yet, but the application of Big Data software algorithms is elevating decision-making precision to a whole new level, creating efficiencies, saving costs or delivering new solutions to important problems. This will be an avatar for some of the best investment opportunities in years to come. Our private firm, Xerion Investments, is investing primarily in what I’ll call “A-AIaaS” plays — Applied Artificial Intelligence-as-a-Service.
After decades of speculation and justifiable anxiety about the social implications for humankind, the AI era is finally here. The Bank of England estimates that 48% of human workers will eventually be replaced by robotics and software automation, and ArkInvest predicts that 76 million U.S. jobs will disappear in the next two decades — almost 10 times the number of jobs created during the Obama years. And it’s more than low-skilled factory workers losing their seats to software intelligent robots. Even Wall Street is being disrupted: cloud-based software technologies, such as Blockchain, are displacing sales/trading and settlements professionals and increasing price discovery and transparency on the sell-side of the Street, while data analytics is helping quantitative trading eliminate the fundamental mis-pricing of securities, formerly the mission and exclusive domain of active mangers.
For instance, one of the companies we invest in uses drones to replace costly work teams driving around in trucks to monitor perimeter security or operations at large worksites like office campuses, mine sites, wind and solar power farms. Considering re-tooling yourself at drone pilot school? Don’t bother, because this company flies drones autonomously with computer vision — algorithms that gather and analyze such detailed data on the specific site environments that the drones will be able to fly themselves around obstacles. Think about what that means for monitoring a working wind turbine, with blades in full motion.
Deep Learning AI isn’t only about cost savings, either — applied data analytics are solving important problems. For example, Xerion is invested in a company that analyzes streaming video feed, applying multiple algorithms (local and cloud-based) to varied object databases to identify specific conditions, objects and people in the video feed. The company can tell merchandizers what brands of clothing teenage concert-goers are wearing at Coachella; what is it that viewers don’t like about TV ads that causes them to change the channel. Cool, and good for business, but there are more purposeful applications: Visual Deep Learning technology will save lives, by precisely identifying suspected terrorists and hidden ordinance in remotely-recorded video feed — a much-needed and truly meaningful safety benefit for our service men and women. Another of our A-AIaaS companies gathers, analyzes and reports air quality conditions local to the user’s hand-held device, navigating people toward a healthier life.
The societal implications of important economic developments, and associated investment opportunities, are two sides of the same coin. I’ve always been equally interested in both, and have anticipated and invested ahead of some of the largest economic developments of the past 30 years, including the global transition away from Socialist economics; China’s economic transition through the stages of industrialization, urbanization and finally consumption-driven growth; the U.S. mortgage and financial crisis & monetary policy-induced recovery of financial markets; and the seismic U.S.-led energy revolution. The social and investment implications of the AI economy look to me even more monumental than any of these prior historic developments.
The accelerating penetration of job-displacing software presents maybe the most serious (and still way under-appreciated) socio-economic challenge to market economies in generations, both in our own country and abroad. For example, China, the world’s largest economy after the U.S., still has over 100 million people working in manufacturing jobs with only 36 robots for every 10,000 workers, versus only 12 million factory workers in the U.S., which in turn has a robot penetration rate of 4.5 times China’s, 164 per 10,000 human workers. Robots have a long way to go and will get there quickly in both countries. How will China’s displaced workers, and consumption as the future engine of the country’s economic growth, be affected?
If fewer people everywhere will be needed for work, and those who will have jobs may work much shorter hours thanks to AI software, how will people earn money to live? We’re starting to hear a lot about this, because entrepreneurs, investors and shareholders of companies will be enjoying epic financial rewards from the AI economy — but what about everyone else?
Prepare for intensifying, potentially de-stabilizing, social tension over winners and losers, income and wealth disparities in the AI economy, and what to do about them to support people and the consumption-driven economies. Of course the news isn’t all bad: applied software technology reduces costs and prices, taking fewer consumption dollars a longer way. But people still need jobs (or will be it be some form of guarantee minimum income from governments from taxing who and how much?), and how will the idle, voluntary or otherwise, stay productively occupied?
On the benefits side, you ask, ‘What’s the trade?’ This isn’t a trade; it’s a multi-year investment. There are always opportunities to buy the publicly traded stock of a lot of great companies like Google, googl Facebook fb and Amazon amzn at the cutting edge of the AI economy, and semi-conductor companies like Nvidia and NXP powering smart software. But the ideas we’re talking about here are so monumental that investment performance needs to be evaluated over years, not days. And, in my view, the most interesting opportunities are in early and growth stage start-ups, private companies cultivating novel and disruptive, if not game-changing, solutions.
A few of the framing assumptions we’re using to pursue some great opportunities of this AI economy:
The software tech is a tool, great investments need commercial applications
I see dozens of presentations and attend many meetings every week with entrepreneurs working developing “proprietary” software algorithms. However, when AI reaches its most sophisticated capabilities, by definition the software is functioning independently and autonomously. What makes it interesting as a business and investment, is how it’s being applied — in what business, for what mission and with what economic benefit? I believe AI algorithms should ultimately be considered a tool— a means like this generation’s equivalent of an Excel spreadsheet, rarely an investment itself.
Only once we satisfy ourselves that the proposed commercial use-case makes sense, do we proceed with the very same diligence we’ve conducted over decades, on questions like: Is there sufficiently capable and robust management team? What’s their market access strategy? Do they understand the time and have the resources to penetrate customers and convert them to revenue before running out of money?
Customers Don’t Want Hardware Only
The software economy is changing the very mission of entire industries, even in the technology industry itself. For example, two of the world’s best-known hardware technology companies, Cisco and Ericsson, recently learned that many customers are shying away from buying hardware and integrating it themselves. Customers now want economic outcomes, usually demonstrable efficiency gains translating to margin improvements. Producing the desired outcomes involves integrating hardware with software and implementation with cloud-based monitoring and process management services. Cisco and Ericsson joined forces to provide “managed service” solutions to their customers; the next generation of A-AIaaS companies are going to do it themselves with integrated solutions and set narratives for defining and penetrating target use-cases. CAPEX-intensive business models are changing so profoundly that even telecommunications networks are moving to “virtualize” essential functions to cloud-based managers, instead of owning and upgrading them as owner/operators—a process appropriately called “network function virtualization.” One of our companies focuses on that specific application; another one detects operational and security anomalies in networked IoT (Internet-of-Things), cloud-based managed services arrangements targeting a range of traditional industrial settings.
Applied artificial intelligence has the making of a great investment.
The most compelling characteristic of A-AIaaS investment opportunity is that, in most cases, the economic benefit is driven by software that’s driven on legacy, or off-the-shelf, hardware racks (there are of course also exceptional hardware innovations, like, revolutionary upgrades of agricultural instruments and processes we’re considering, which haven’t been updated in 40 years). Almost no CAPEX; A-AIaaS arrangements are mostly OPEX — efficiencies for hire — benefits dropping right to operating margins and bottom line. The AI revolution is actually about to transform corporate margins beyond anything we’ve seen in recent history, and without the need for customers to carry capital charges on their balance sheets, because there are almost none: Compare the R & D amortization of the investment in a crew of smart young software engineers in hoodies, to the Billions of R & D dollars spent developing blockbuster drugs.
A-AIaaS raises important economic policy challenges, but it also looks to be the most exciting and widespread potential capital efficiency and ROI opportunity seen in years.
Daniel J. Arbess is the founder CEO of Xerion Investments, an investor and policy analyst recognized for his prescient calls on some of the largest developments of the past 30 years; and a cofounder of No Labels, the bipartisan policy organization.