Mobile Is Eating the World, 2016

Mobile Deployment and Industry Scale

• Global mobile connectivity is at the 50% mark, with 5 billion people owning mobile phones and 2.5 billion using smartphones, a figure projected to reach 5 billion.
• Smartphone adoption is surpassing the PC curve; while there are 1.5 billion PCs globally, mobile usage is accelerating toward total population coverage.
• Technology adoption follows an S-curve where the "creation phase" (platform wars, technology debates) is transitioning to a "deployment phase" focused on scaling and maturity.
• In the deployment phase, platform winners (Google, Apple, Facebook, Amazon) are established, and technology is becoming a commodity with incremental improvements.
• GAFA (Google, Apple, Facebook, Amazon) are approximately three times larger in revenue than the previous dominant "Wintel" (Microsoft, Intel) era companies.
• GAFA companies possess ten times the scale of Microsoft and Intel during the 1990s growth period and now constitute the five largest companies by stock exchange valuation.
• Annual capital expenditure for major tech giants has risen from $1 billion in 2000 to $34 billion in 2015.
• The combined workforce of these dominant tech firms exceeds half a million employees, representing a tenfold increase in headcount compared to the past.
• These companies have evolved from tech-industry giants to dominant economic entities, competing across diverse sectors including retail, content, and hardware.
• Apple ranks as the 10th largest retailer globally with $53 billion in sales, while Amazon uses content merely as a lever for Prime subscribers and product sales.
• Netflix holds the fourth-largest production budget in the USA, and Intel's chip-making business has become a side project for tech giants building custom silicon (FPGAs, ASICs, SoCs).
• Google's 2010 executive dismissed Netflix as a threat comparable to the "Albanian army," a sentiment that has since reversed as tech companies fundamentally impinge on traditional industries.

Machine Learning and Data Shifts

• Machine learning breakthroughs in 2012 shifted industry focus from rule-based programming to data-driven neural networks, rendering previous decades of AI research obsolete.
• Image recognition error rates dropped from 28% to 7%, while speech recognition error rates fell from 26% to 4%, enabling previously impossible computational tasks.
• Frederick Jelinek's adage applies: "Every time I fire a linguist, the performance of the speech recognizer goes up," signaling a shift from human-defined rules to automated data analysis.
• Google applied AlphaGo's system to data center cooling, achieving a 15% energy saving after decades of failed manual optimization attempts.
• Google CEO Sundar Pichai announced a strategic pivot from "Mobile First" to "AI First," refocusing the entire tech industry on machine learning capabilities.
• Machine learning models can now be deployed at the edge on low-cost devices (e.g., $10-$20 widgets) using image sensors, enabling vision capabilities in non-traditional hardware.
• The global camera market is shrinking, but image sensors have become universal components embedded in almost all electronic devices.
• Cloud providers (Google, Amazon, Microsoft, IBM) are abstracting machine learning complexity into accessible platforms, allowing developers to integrate speech and image recognition via APIs.
• Machine learning is currently in the early "creation phase," characterized by invention frenzies, debates on winning strategies, and high potential for future growth.

New Computing Paradigms and Interaction

• Smartphone apps now account for 60% of all online time in the USA, with Facebook absorbing 15-20% of that time, creating high barriers to entry for new apps.
• The smartphone model has replaced the PC interface by integrating touch, tilt, sensors, and imaging as primary inputs, removing the need for keyboards and mice.
• Live video apps have emerged by assuming high-end smartphone penetration (1 billion devices), unlimited home Wi-Fi, and constant connectivity, breaking previous interface assumptions.
• Frictionless devices like Amazon Echo and Snapchat Spectacles operate by unbundling traditional apps into hardware that eliminates administrative steps like charging, drivers, and app selection.
• The strategic direction of computing is defined by the removal of user questions and choices (e.g., passwords, file locations), creating a strong incentive for platforms to control the answer.
• Human-computer interaction is evolving from direct physical manipulation of data (pinning paper) to direct physical interaction (touching, looking, talking) and finally to augmented reality (gestures).

E-Commerce Transformation

• The internet is replicating its transformation of media within the retail sector, unbundling physical bundles like malls and stores into new digital aggregators.
• E-commerce currently represents 10-12% of US retail revenue; Amazon accounts for at least 2% (or up to 6% including marketplace revenue).
• E-commerce has proven superior at logistics than at demand generation, successfully selling known products but lacking systematic solutions for discovering new items.
• Amazon functions as "Google for products," yet there is no equivalent "Facebook for products" or "BuzzFeed for products" to suggest items based on user taste.
• The channel of purchase fundamentally dictates consumption; changing the channel changes what consumers buy, rendering the channel itself the product.
• "Soap as a Service" models (e.g., Dash buttons, Alexa) disintermediate retailers, brand builders, and ad agencies by automating recurring purchases without brand decision-making.
• Machine learning and imaging are expected to revolutionize demand generation by analyzing personal environments (e.g., a photo of a living room) to suggest products based on personal taste.
• Retail is a $20 trillion opportunity ripe for disruption by software, machine learning, and mobile, similar to how software disrupted consumer media and pay TV.

Automotive Industry Disruption

• The automotive sector is undergoing a bifurcated transformation: the shift to electric vehicles (EVs) and the shift to autonomy, which are distinct but related paths.
• Electric vehicles destabilize the industry by removing the engine and transmission, replacing complex proprietary systems with commodity batteries and motors (10x fewer moving parts).
• Value in the EV transition is shifting from hardware manufacturing to software, requiring a new skill set and allowing tech giants with massive capital reserves (Apple: $237B, Google: $73B) to compete.
• Battery costs must drop from over $200 per kWh to approximately $100 per kWh to achieve cost parity with gasoline engines.
• Autonomy levels range from Level 1 (cruise control) to Level 5 (fully autonomous without steering wheels); Level 5 autonomy is projected 5 to 10 years away.
• Future competitive moats in autonomy will rely on network effects from driving data, high-density 3D mapping, and on-demand service density rather than physical hardware.
• Full autonomy is expected to have second-order effects that surpass the removal of the horse in the "horseless carriage" analogy, fundamentally altering cities, housing, logistics, and insurance.
• Historical predictions from 1900 often fail to account for societal changes beyond the technology itself (e.g., predicting flying cars without predicting Walmart or freeways).
• The most significant impacts of autonomous vehicles will likely be in the elimination of parking, reduction of congestion, and the restructuring of commercial real estate and trucking.