Palo Alto – June 10, 2019 – Amazon held an outstanding TED-like Artificial Intelligence / Machine Learning conference last week in Las Vegas – Re:MARS – which brought together 3,000 leaders in applied AI.  Over 100 experts were featured – from Amazon itself, as well as from science, academia and business – who shared their future thinking.  Re:MARS demonstrated why Amazon has such a big head start in applied AI.  The company is using AI to reinvent itself at scale – infusing AI through virtually all its various businesses – particularly eCommerce, AWS Services, Alexa, Amazon Go, Prime Air Drones and Amazon Robotics.

As Woodside Capital Partners’ Sector Lead in AI/ML, I looked at Re:MARS through the lens of how development of AI/ML technology will fuel financing, mergers and acquisitions.  The conference underscored how massive the AI tidal wave will be – much industry growth lies ahead, and part of that industry growth will be the acquisition of many AI/ML companies (currently there are 7,000 private AI/ML companies). Near-term consolidation is inevitable.

Amazon – like Google, Microsoft and certain others – has spent tens of billions of dollars building out AI-related infrastructure and services.  The company claims to offer the broadest feature set for AWS customers to deploy commerce, geography-related, voice, robotics, prediction, language, vision and other advanced products.  However Amazon is frustrated because companies are not deploying AI/ML quickly enough.   

From an acquisition perspective, while the largest tech companies have an enormous advantage in AI because they own deep oceans of data, large engineering talent pools, and vast pools of capital, they also still have big deficiencies in their AI/ML portfolios – and this will fuel acquisitions.  For example, Amazon itself bolstered its own robotics capabilities through its purchase of Kiva Systems ($775M headline price), the Massachusetts-based company that builds mobile robotic fulfillment systems, and through its purchase of Canvas Technology, the Colorado-based firm that is developing a fully autonomous cart system.

Highlights and raw session notes from Amazon Re:MARS follows:

• Jeff Bezos fireside chat:
  • On disrupting his own organization: “When considering future strategy for Amazon, one of the most interesting questions we ask is – what things should we not change in the next 10 years?  Like in Amazon’s e-commerce business – we feel confident that 10 years from now people will still want low prices, fast shipping and high quality … and we find that usually the static unchanging factors like this are centered around customer needs. But factors like technology, competitors, market factors – those factors are more dynamic and that’s where we should focus our thinking around what should change.”
  • On knowing when it’s time to end a failing initiative: “I don’t like to throw in the towel.  It’s hard.  What I do is look around the room – and when the last high-judgement champion – say someone I’ve known for decades – throws in the towel, that’s when I know it’s time to stop.”
  • On the space company Blue Origin: “We are looking to return to the moon.  It is a resource-rich place.  There is ice which can make propellent, gravity is 6 times less, we know a lot more about the moon now that we didn’t know during the lunar program.  The vision is that we need to move things into space.  We need to protect earth.  We can move the industry to space to protect our planet.  This isn’t for our generation.  It’s for our grandchildren.  We are just starting.  To get infrastructure is always expensive.  When we started Amazon, the transportation system and payment systems were in place – USPS and credit card companies.  You cannot start an interesting space company from your dorm room at this point, like for example, Facebook was started.  My goal with Blue Origin is to build infrastructure.”
  • On what he would be doing if Amazon had failed: “If Amazon had not worked out, I would probably be a very happy software engineer.”
• How AWS is making a push to get enterprise customers to build more AI applications. 
  • Recent McKinsey study: half of American companies have not deployed any AI applications, and another 30% have only deployed it in very limited ways.  Only 3% have mandated AI has a critical priority and deployed it enterprise-wide.
  • Werner Vogels, Amazon CTO: “AWS has a full suite of products for AI & Machine Learning.  It has never been easier to deploy AI – We are giving you the tools, so you should go out and build”.  
  • Amazon AWS has built a robust suite of tools for companies that want to deploy AI and ML. Literally hundreds of features available. Amazon SageMaker has features around Vision, Speech, Language, Recommendations, Forecasting and more.  Vogels showed a particularly cool application of Amazon SageMaker combined with RFID tag sensors in NFL players’ shoulder pads enabling next-gen stats on ESPN football broadcasts such as speed, separation, distance, and completion probability %.  Much richer viewing experience for football fans.  Clean data – essential to AI – Amazon has machine learning tools to automatically clean data.
  • Andrew Ng – founder Google Brain, former CTO Baidu, Professor Stanford, founder Coursera on how to implement AI in an enterprise that doesn’t currently have AI:
    1. Start small
    2. Automate tasks not jobs
    3. Combine AI Experts and Subject Matter experts in a company … select 6 projects … fund 2 or 3
       • Find the right use cases – in the software industry it’s easy to forget how businesses in other industries are run on tribal knowledge
• How Amazon is innovating around drones
  • A drone was unveiled at the conference that can fly up to 15 miles roundtrip to a customer and deliver packages up to 5 pounds – which accounts for 75%-90% of all packages delivered via Amazon. Big focus around safety – the Amazon drone will be able to avoid other aircraft, fly on its own, produce safe, predictable behavior “in every situation”, sense and avoid moving and static objects using visual, thermal, ultrasonic (important because, for example, fluffy dogs are invisible to sonar) will fly below 400 feet.  Wire detection is a huge challenge – electric, phone, laundry – and they have tech that can sense wires
• How Amazon is innovating around Alexa
  • Amazon Echo launched 2014 and is now available in 80 countries and 40 languages.  60k smart home devices are now compatible with Alexa, and Alexa has 90,000 skills
  • For their Alexa platform, Amazon is focused on four AI/ML pillars:
    1. Enhance and assure customer trust – full transparency.  Alexa privacy hub recently launched.  Can delete what Alexa just heard.
    2. Make Alexa smarter as you use Alexa via self-learning.  Unsupervised speech learning.  Learn directly from customers. 
    3. Make Alexa more proactive – context of who, what, when, where.  Alexa will deliver hunches – “I think you left the garage light on”. Or “Possible glass break heard” or “Possible CO2 alarm heard”.
    4. Enabling multiple requests:
       • Play song and turn on the lights
       • “Alexa start cleaning” instead of “Alexa start Rumba cleaning”
       • Alexa when do I have to go to Mom’s house?
  • Generally anticipating customers’ latent goals
    • Not just “what movies are playing tonight” but will anticipate dinner, transportation.  Goal is seamless interaction.
• On Robots: 
  • Colin Angle – Founder/CEO of iRobot – “Autonomous robots have been built and they are great.  We have built and sold 20 million autonomous robot vacuum cleaners.  But autonomy is not intelligence.  If robotics is going to take the next leap forward, we need to figure out how to make robots actually intelligent.”
  • Teaching robots to grasp objects is a big challenge.  Scientists and engineers have been working on it for years.
• On Product Recommendations and AI/ML
  • AI/ML drives significant revenue at Amazon.com via product recommendations – eg – “customers who bought this also bought …” and other forms of recommendation.  Product recommendations aren’t always easy.  Factors considered:
    • Scale – 100s of millions of customers, 100s of millions of products
    • Latency – recommendations need to be very fast, immediate
    • Dimensionality – challenge is that what customers care about is different and changeable, especially challenging with diverse product categories
    • Localization – inventory/availability of products; cultural norms & nuances; languages; logistical support
    • Subjectivity – everyone perceives their recommendations differently; there is no ground truth
    • Evaluation – success is ultimately measured by customer satisfaction in the real world
    • Intent – Hard to understand shopping intent due to varying intent across people, Intent changes over time, one person may have multiple intents.
  • In the US, 60% of dollars spent goes to non-durable goods – food & beverages, toiletries, apparel, personal care, beauty, etc.  These are somewhat easy to predict – when someone will need toilet paper or dog food based on frequency of past purchases.   But certain categories – fashion for example – are very challenging to predict.
• On  reinventing manufacturing through AI/ML – Guido Jouret, Chief Digital Officer, ABB
  • ABB is a 136-year-old Swiss industrial company producing heavy electric equipment.  Runs 50 factories.
  • Focus in recent decades has been on Automation – analyze sense act.   Focus shifting to Autonomous – machines that can do more and act independently
  • Essential to improve efficiency of factories – from 1990-2010: 4% efficiency improvement per year in factories … 2011-2018 only 1% efficiency improvement per year.  AI/ML is the key.  50% of the world’s electricity is consumed by motors – many in factories.
  • Vision for Industry 4.0 manufacturing:
    • Intelligent sensors enabling process automation
    • Mobile Robots – GPS/5G
    • Better logistics & shipping
  • AI systems are about task completion – AI isn’t good at what people do.  Intuition, creativity, expertise
  • World population will grow by 2B between now and 2050, and we aren’t prepared:
    • 50% of food lost in transportation or thrown away
    • Up to 20% of water lost to leaky pipes
    • 20% of energy used moving water
    • 6% of energy lost in transmission
    • We need to move to renewable fuel
  • Electric car penetration is about to take off.  US is at 2% electric, China is at 5% electric … it takes about the same amount of time to go from 0% to 1% penetration of a new product as it does to go from 1% to 50%.  We are at an inflection point.
  • Farming has a specific challenge – the average age of farmers in Japan is 67.  Same issue in rest of world. 

For more information please contact:

Kelly Porter
Lead Managing Partner
Woodside Capital Partners International LLC
[email protected]  

About Woodside Capital Partners
Woodside Capital Partners is a global, independent investment bank that delivers world-class strategic and financial advice to emerging growth companies in the technology and healthtech sectors. With a strong track record in M&A, strategic partnerships and private placements, Woodside Capital Partners has been providing worldwide investment banking services since 2001 with leading domain experience in software, Internet services, electronic communications and materials, and healthcare. Woodside Capital Partners is headquartered in Silicon Valley. Securities offered through Woodside Capital Securities LLC, member FINRA/SIPC. For more information, please visit www.woodsidecap.com . NOTICE: No information herein is a recommendation that any particular individual should purchase or sell any particular security in any amount or at all, and is not a solicitation of any offer to purchase or sell from or to any particular individual.