An AI supply chain starts with the sourcing of data, which is produced by consumers. As this data gets stored on hardware, it goes through a first refinement process via software. Then it’s further refined, and repackaged by algorithms, and stored in data centers, which work as the fulfillment centers.
Google’s cloud is an interesting example of how information flips the supply chain upside down.
As pointed out in vertical integration, the physical supply chain gets flipped upside down, when we look at it from a data/information standpoint.
A classic supply chain moves from upstream to downstream, where the raw material is transformed into products, moved through logistics and distributed to final customers. A data supply chain moves in the opposite direction. The raw data is “sourced” from the customer/user. As it moves downstream, it gets processed and refined by proprietary algorithms and stored in data centers.
To truly appreciate those differences, let’s start with one of the most interesting of the physical supply chains: coffee.
From coffee beans to data farms
The coffee supply chain is among the most interesting, as it’s quite complex and it goes through a cycle of growing beans, harvesting, drying, packing, bulking, blending, roasting, labeling, packaging, distributing, and selling.
Throughout this process, there are many players involved globally, at each stage. Perhaps wherein the growing and harvesting of coffee beans, countries like Brazil and Colombia play a leading role.
Yet, as the bean goes through a process of refinement, it goes through several parts of the chain, and only toward the end, there is the process of roasting, labeling, packaging, and distribution.
Most of the economic value of this supply chain is skewed toward the end. As we get closer to roasting, labeling, distributing, and retailing, that is where most of the economic output is captured.
In other words, of the overall price paid by customers in the shops for their nice double espresso, most of the revenues go toward covering up for the expenses to run the shop/rent, the staff, tax, and profits.
To gain a bit of context, as the Financial Times evaluated on a 2.50-pound cup of coffee, 25 pennies go to the shop as a profit, and only 10 pennies go to the overall coffee chain, and about a penny goes toward the farmer.
If we use this analogy for the world of AI supply chains, the farmer is no longer in a plantation in South America. But anywhere tapping with her/his finger on a 4.5 inches smartphone.
Raw materials are sourced by consumers
In an AI supply chain it all starts from consumers. They are the growers of the raw materials (raw data) that will serve as the basis for the whole supply chain.
It’s worth to point out, that, as in a coffee supply chain, where the beans are grown by farmers, which are the ones capturing less of the individual economic value from the supply chain.
In an AI supply chain, consumers are like farmers, and they also are the ones that gain the least in terms of economic value from the overall supply chain.
Hardware devices become the harvesting facilities
The raw data gets gathered, harvested and collected through physical devices, which are the most proximate object to the consumer.
As the raw data becomes available, hardware becomes the harvesting facility in the AI supply chain.
Software and operating systems become the harvesting machines
Software, operating systems and everything else that is in between the physical device and the company’s algorithms become the harvesting machines, ready to sort the data, that will go through the industrial machineries of the AI supply chain.
Algorithms are the industrial machineries for data
As this data, partly filtered by the software side will go through a process of industrial refinement, algorithms will play a key role in refining, processing, and packaging of the data for several scopes.
In that sense, the data moves in two directions. On the one hand, it will move toward consumers to improve the services they get for free. On the other hand, it will move into the proprietary technology stack of the company, ingrained in its monetization machinery to generate profits.
Before it can move in those two directions, though, it will need to be stored within its data centers.
Data centers as fulfillment facilities
As the data goes through the data centers, it gets stored, and it moves in many directions. Back to consumers in the form of free services and toward the monetizaiton machinery, where the data processed refined, and continously reprocessed become the core servive the company offers on the market.
Perhaps, as Google highlights “Our data centers keep all of Google’s products and services up and running around the clock and around the world. Whenever you access Gmail, edit your documents, or search for information on Google, you’re using one of our data centers and have the power of a supercomputer at your fingertips.“
Key takeaways
The consumer as the farmer sources the raw data, and it gets back only a fraction of the overall economic value.
Most costs go back to data centers, power sourcing, profits, and organizational costs.
Algorithms work in two directions, by refining data to offer free services to consumers. And by creating the premises for the monetization machinery to work at the best.
Hardware, as the most proximate thing to the consumer, becomes the harvesting facility.
Data centers become the fulfillment facilities moving refined and repackaged data in two directions. Toward consumers in the form of free services and toward businesses in the form of advertising or premium services.
Is this a permanent design for AI supply chains? Not necessarily. Yet that has become the predominant design for new dominant media companies (Google, Facebook).
Key Highlights
AI Supply Chain Overview:
An AI supply chain begins with data sourcing from consumers.
Data goes through refining processes via software and algorithms.
Algorithms process and repackage data, storing it in data centers.
Data centers act as fulfillment centers, serving both consumers and monetization.
Comparison to Physical Supply Chains:
The traditional supply chain moves from raw materials to end products.
In the data supply chain, raw data comes from consumers and moves downstream through refining and processing.
Consumer Role in AI Supply Chain:
Consumers are equivalent to farmers in the data supply chain, providing raw data.
Similar to coffee farmers, consumers capture a relatively small portion of economic value.
Data Collection and Harvesting:
Hardware devices act as harvesting facilities, collecting raw data from consumers.
Software, operating systems, and intermediary elements function as harvesting machines.
Role of Algorithms:
Algorithms play a pivotal role in refining, processing, and packaging data.
Data moves in two directions: enhancing consumer services and feeding monetization systems.
Data Centers as Fulfillment Facilities:
Data centers store and manage processed data, serving various purposes.
Data is directed back to consumers through free services and to businesses for monetization.
Key Players and Costs:
Consumers contribute raw data but capture a fraction of the overall value.
Most costs are associated with data centers, power, profits, and organizational aspects.
Algorithm Direction and Hardware Role:
Algorithms refine data for free consumer services and provide the foundation for monetization.
Hardware, as the closest element to consumers, acts as the initial harvesting facility.
The supply chain is the set of steps between the sourcing, manufacturing, distribution of a product up to the steps it takes to reach the final customer. It’s the set of step it takes to bring a product from raw material (for physical products) to final customers and how companies manage those processes.
A classic supply chain moves from upstream to downstream, where the raw material is transformed into products, moved through logistics and distribution to final customers. A data supply chain moves in the opposite direction. The raw data is “sourced” from the customer/user. As it moves downstream, it gets processed and refined by proprietary algorithms and stored in data centers.
Distribution represents the set of tactics, deals, and strategies that enable a company to make a product and service easily reachable and reached by its potential customers. It also serves as the bridge between product and marketing to create a controlled journey of how potential customers perceive a product before buying it.
A distribution channel is the set of steps it takes for a product to get in the hands of the key customer or consumer. Distribution channels can be direct or indirect. Distribution can also be physical or digital, depending on the kind of business and industry.
In business, vertical integration means a whole supply chain of the company is controlled and owned by the organization. Thus, making it possible to control each step through customers. in the digital world, vertical integration happens when a company can control the primary access points to acquire data from consumers.
Horizontal integration refers to the process of increasing market shares or expanding by integrating at the same level of the supply chain, and within the same industry. Vertical integration happens when a company takes control of more parts of the supply chain, thus covering more parts of it.
By definition, a horizontal market is a wider market, serving various customer types, needs and bringing to market various product lines. Or a product that indeed can serve various buyers across different verticals. Take the case of Google, as a search engine that can serve various verticals and industries (education, publishing, e-commerce, travel, and much more).
A vertical or vertical market usually refers to a business that services a specific niche or group of people in a market. In short, a vertical market is smaller by definition, and it serves a group of customers/products that can be identified as part of the same group. A search engine like Google is a horizontal player, while a travel engine like Airbnb is a vertical player.
When entering the market, as a startup you can use different approaches. Some of them can be based on the product, distribution, or value. A product approach takes existing alternatives and it offers only the most valuable part of that product. A distribution approach cuts out intermediaries from the market. A value approach offers only the most valuable part of the experience.
Backward chaining, also called backward integration, describes a process where a company expands to fulfill roles previously held by other businesses further up the supply chain. It is a form of vertical integration where a company owns or controls its suppliers, distributors, or retail locations.
A market type is a way a given group of consumers and producers interact, based on the context determined by the readiness of consumers to understand the product, the complexity of the product; how big is the existing market and how much it can potentially expand in the future.
Psychosizing is a form of market analysis where the size of the market is guessed based on the targeted segments’ psychographics. In that respect, according to psychosizing analysis, we have five types of markets: microniches, niches, markets, vertical markets, and horizontal markets. Each will be shaped by the characteristics of the underlying main customer type.
According to the book, Unlocking The Value Chain, Harvard professor Thales Teixeira identified three waves of disruption (unbundling, disintermediation, and decoupling). Decoupling is the third wave (2006-still ongoing) where companies break apart the customer value chain to deliver part of the value, without bearing the costs to sustain the whole value chain.
Disintermediation is the process in which intermediaries are removed from the supply chain, so that the middlemen who get cut out, make the market overall more accessible and transparent to the final customers. Therefore, in theory, the supply chain gets more efficient and, all in all, can produce products that customers want.
Reintermediation consists in the process of introducing again an intermediary that had previously been cut out from the supply chain. Or perhaps by creating a new intermediary that once didn’t exist. Usually, as a market is redefined, old players get cut out, and new players within the supply chain are born as a result.
As startups gain control of new markets. They expand in adjacent areas in disparate and different industries by coupling the new activities to benefits customers. Thus, even though the adjunct activities might see far from the core businessmodel, they are tied to the way customers experience the whole businessmodel.
The bullwhip effect describes the increasing fluctuations in inventory in response to changing consumer demand as one moves up the supply chain. Observing, analyzing, and understanding how the bullwhip effect influences the whole supply chain can unlock important insights into various parts of it.
Dropshipping is a retail business model where the dropshipper externalizes the manufacturing and logistics and focuses only on distribution and customer acquisition. Therefore, the dropshipper collects final customers’ sales orders, sending them over to third-party suppliers, who ship directly to those customers. In this way, through dropshipping, it is possible to run a business without operational costs and logistics management.
Consumer-to-manufacturer (C2M) is a model connecting manufacturers with consumers. The model removes logistics, inventory, sales, distribution, and other intermediaries enabling consumers to buy higher quality products at lower prices. C2M is useful in any scenario where the manufacturer can react to proven, consolidated, consumer-driven niche demand.
Transloading is the process of moving freight from one form of transportation to another as a shipment moves down the supply chain. Transloading facilities are staged areas where freight is swapped from one mode of transportation to another. This may be indoors or outdoors, depending on the transportation modes involved. Deconsolidation and reconsolidation are two key concepts in transloading, where larger freight units are broken down into smaller pieces and vice versa. These processes attract fees that a company pays to maintain the smooth operation of its supply chain and avoid per diem fees.
Break bulk is a form of shipping where cargo is bundled into bales, boxes, drums, or crates that must be loaded individually. Common break bulk items include wool, steel, cement, construction equipment, vehicles, and any other item that is oversized. While container shipping became more popular in the 1960s, break bulk shipping remains and offers several benefits. It tends to be more affordable since bulky items do not need to be disassembled. What’s more, break bulk carriers can call in at more ports than container ships.
Cross-docking is a procedure where goods are transferred from inbound to outbound transport without a company handling or storing those goods. Cross-docking methods include continuous, consolidation, and de-consolidation. There are also two types of cross-docking according to whether the customer is known or unknown before goods are distributed. Cross-docking has obvious benefits for virtually any industry, but it is especially useful in food and beverage, retail and eCommerce, and chemicals.
The Toyota Production System (TPS) is an early form of lean manufacturing created by auto-manufacturer Toyota. Created by the Toyota Motor Corporation in the 1940s and 50s, the Toyota Production System seeks to manufacture vehicles ordered by customers most quickly and efficiently possible.
Six Sigma is a data-driven approach and methodology for eliminating errors or defects in a product, service, or process. Six Sigma was developed by Motorola as a management approach based on quality fundamentals in the early 1980s. A decade later, it was popularized by General Electric who estimated that the methodology saved them $12 billion in the first five years of operation.
Scientific Management Theory was created by Frederick Winslow Taylor in 1911 as a means of encouraging industrial companies to switch to mass production. With a background in mechanical engineering, he applied engineering principles to workplace productivity on the factory floor. Scientific Management Theory seeks to find the most efficient way of performing a job in the workplace.
Poka-yoke is a Japanese quality control technique developed by former Toyota engineer Shigeo Shingo. Translated as “mistake-proofing”, poka-yoke aims to prevent defects in the manufacturing process that are the result of human error. Poka-yoke is a lean manufacturing technique that ensures that the right conditions exist before a step in the process is executed. This makes it a preventative form of quality control since errors are detected and then rectified before they occur.
A Gemba Walk is a fundamental component of lean management. It describes the personal observation of work to learn more about it. Gemba is a Japanese word that loosely translates as “the real place”, or in business, “the place where value is created”. The Gemba Walk as a concept was created by Taiichi Ohno, the father of the Toyota Production System of lean manufacturing. Ohno wanted to encourage management executives to leave their offices and see where the real work happened. This, he hoped, would build relationships between employees with vastly different skillsets and build trust.
Jidoka was first used in 1896 by Sakichi Toyoda, who invented a textile loom that would stop automatically when it encountered a defective thread. Jidoka is a Japanese term used in lean manufacturing. The term describes a scenario where machines cease operating without human intervention when a problem or defect is discovered.
The andon system alerts managerial, maintenance, or other staff of a production process problem. The alert itself can be activated manually with a button or pull cord, but it can also be activated automatically by production equipment. Most Andon boards utilize three colored lights similar to a traffic signal: green (no errors), yellow or amber (problem identified, or quality check needed), and red (production stopped due to unidentified issue).
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
