I’ve had the opportunity to engage in a significant manner and duration with many leading companies over the last decade regarding artificial intelligence (AI), their corporate strategies, and of course systems thereof, which is our domain. Since enterprise-wide AI systems are synonymous with strategic imperatives, observations gained in this process are rare if not unique, providing interesting and invaluable insights.
The situation and capabilities of companies we’ve dealt with vary across a wide spectrum, ranging from a very few at the top tier in strategic competency, to a few more in a fatal tailspin. The vast majority of companies are somewhere in-between, representing the bulk of companies that need help that are also in a position to receive it.
The following represent AI strategies that have emerged over the past decade. Some have proven to be much more successful than others.
The mastodon
This is typically the market leader in a century-old industry that benefits financially from entrenched relationships with governments and industry. Mastodons are justifiably paranoid about the risk of an extinction event. They feel massive weight on their shoulders from a long rich history and large family reliant on them as head of their ecosystem. They typically apply their financial strength to outspend competitors combined with other brute strengths such as lobbying. Most are claiming to have successfully transformed into a big tech or successful unicorn. There is just one problem with that strategy, which is that it isn’t true. Despite massive hiring and spending, with some success of course, they are neither. Their priors embedded in DNA are different, as are inherited strengths and weaknesses. They are simply a different species with different opportunities, risks, advantages, and cultures.
Mastodons are much more skilled at financial engineering than they are in technology, but understand the need to invest to attract talent and provide play boxes to keep people occupied, hence all the talk about innovation. Just don’t expect much change. Mastodons are usually considered strategic to large national economies, are too big to fail, and so may be able to power on by sheer inertia until the next killer asteroid hits earth. Since so few mastodons are left and they typically don’t think they need anyone’s help, I won’t spend additional time on them except to make one point.
The auto industry has invested the most capital in AI to date, and did so earlier than most. Although most of the market leaders started with the mastodon strategy, all but one auto giant has since determined that they weren’t too big to fail, they couldn’t do it alone, and needed help, so they have formed a variety of interesting partnerships. These companies are being forced by technology and disruption to become transformers. The same may or may not be true for other industry leaders. For now, the mastodon strategy appears to be working only for a very few (single digits). Some apparent early success may actually be an effective stalling technique rather than long-term strategy (we should find out within the next five years).
The transformers
I’ve identified three types of transformer companies. The most dynamic and successful are true transformers typically led by a management team and board that well understand the existential risks facing their company. They are rare, but can be found in most industries. These scenarios are massive turn-around efforts, preferably before the companies are facing serious disruption. Transformations don’t work terribly well under severe financial stress. True transformers don’t produce many press releases on the topic as they are focused on the heavy lifting of replacing outdated business models. They depend on results to move share price, not marketing fluff. True transformers tend to make good partners for serious startups due to similar cultures and needs. Those attempting to transform into something different for survival must take greater risks. Management and owners usually have more personal skin in the game. A lot is at stake with true transformers. If they succeed in the transformation, they may become industry leaders. More often than not they don’t succeed and are then forced to rely on investment bankers to find a financial solution in M&A.
Partial transformers
The most common type of transformer is partial, representing the majority of the Fortune 500. Partial transformers enjoy good solid business units that are performing well and are not facing existential risk in the foreseeable future. Think of a leading national restaurant or hotel company—not yet facing the risks of auto, retail, publishing, etc. Partial transformers are a great place to advance one’s careers, particularly as a CXO in mid-career and raising a family. The risks are low and manageable, compensation is good, and one has a reasonably good chance of having a life outside the office and airports. As the name implies, partial transformers do have risk and opportunity, so they are actively experimenting, do some strategic partnering and venture investing, and usually acquire a few companies. Partial transformers usually do well with AI projects with limited goals. They can make great customers due to sufficient financial strength, stability and scale.
Fake transformers
Fake transformers remind me of the old low budget thriller movies where the monsters really don’t look real to anyone. It is almost as if everyone is expected to understand this is just all fantasy theatre, but they hope you buy their stock and products anyway. Fake transformers invest heavily in marketing and PR, buybacks, management compensation packages, and even technology, but underperform their peers and the market. Fake transformers suffer from failed leadership in the CEO and board.
Intelligence agencies
While all companies perform competitive intel, some are more intelligence gatherers than product or service providers. Intel gatherers became quite problematic in AI systems. They are essentially impossible to work with because their intentions are not what they claim. They may act as your best friend to elicit deeper intel, but true intentions typically include your demise. Although this model is obviously not successful as an AI implementation strategy, it can be an effective business model. The company may find an opportunity, partner, or acquirer, or may broker the intel as it may be more valuable than their company. I view this AI strategy as temporary at best because it creates so many enemies.
Vertical maestro
One of the most successful AI strategies to date for incumbent companies has been the vertical maestro. These are usually the market leader in a large specialty industry that is sufficiently technical to require a strong team with consistent investment over decades. Think specialty manufacturing with their own factories and dealership networks. Some of these companies face serious risk from AI combined with other factors, such as emerging competition, political pressure, and significant debt. The tech industry also has many vertical maestros, such as Salesforce. They acquired one of our early peers and have increasingly built their business around AI functionality. Healthcare also has many vertical maestros, though healthcare has been much slower to adopt strategic AI.
These companies were in a good position to leverage machine learning projects and integrate into their existing product lines. However, vertical maestros tend to be optimized in price relative to debt level—sometimes priced higher than the market can withstand. In some cases their customers are seeking alternatives out of necessity. Customers may only have one or two choices when they need a full menu, perhaps even to survive. Therein lies the challenge for vertical maestros—by further consolidating their grip on markets, they may actually be increasing risk of displacement. Vertical maestros make good candidates for lateral moves into other industries, whether through strategic partnerships, investment, acquisition, or creating a new business from scratch.
Platform conglomerate
Amazon is of course the ultimate example of a platform conglomerate, and a fairly effective one at that. However, many platform conglomerates exist, particularly since the commercialization of the Internet, though other forms have existed throughout history. Ecommerce platforms are inherently favorable to conglomeration. Once the network is established, scale is expanding and the company is break-even, expansion into new products and lines of business are fairly easy. Native network companies also have a technical advantage over incumbents due to the lack of legacy IT systems and cultures. Anyone building a new platform today would be foolish not to plan for incorporating machine learning, and would be wise to partner with, or if sufficiently funded, build a complete AI system. Uber is a good example of the latter—they acquired a startup that became their AI lab, which much of the future business was built around, and have since expanded into other areas.
One of the main challenges for platform companies is to resist excessive conglomeration and remain focused on core areas that are profitable and sustainable. I experienced this myself in our lab and incubator in the 1990s when we operated small platforms by comparison to today’s leaders. It was all-too easy to expand into new areas online. Difficult to resist the temptation no matter how wise.
Authentic business builders
These are the most successful new companies in AI now numbering in dozens if not hundreds. Most are specialists such as drug development ventures or autonomous vehicles. A few like my company are focused on universal AI systems. Alexa is a large company example of a universal system even if very different than ours. Tesla is an example of a rapidly growing authentic business builder that integrated AI and became a vertical maestro.
Authentic business builders tend to attract the most driven teams. They are exceptionally mission-oriented and very focused on purpose such as curing disease, preventing crises, space travel, and/or sustainability, whether environmental, economic or as I see it – both. They also tend to be where most of the new wealth is created. I expect future pure play market leaders to emerge from this group. If this doesn’t occur within the next couple of years it is a clear sign of market failure due to the big tech oligopoly, which would cause massive damage to the U.S. economy and future, calling for even more radical antitrust intervention than currently contemplated.
Hybrids
Some of the most successful AI companies are hybrids of the various species. Alphabet for example has an emerging mastodon in Google that is also a platform conglomerate, and owns other companies that are authentic business builders and true transformers. Google is also one of the largest intel gathering networks on the planet. Microsoft, Apple and others have examples in their companies as well that meet the criteria of each of these species, including the occasional fake transformer. Large incumbents should probably employ a hybrid of each type on a selective basis.
The following represent AI strategies that have emerged over the past decade. Some have proven to be much more successful than others.
The mastodon
This is typically the market leader in a century-old industry that benefits financially from entrenched relationships with governments and industry. Mastodons are justifiably paranoid about the risk of an extinction event. They feel massive weight on their shoulders from a long rich history and large family reliant on them as head of their ecosystem. They typically apply their financial strength to outspend competitors combined with other brute strengths such as lobbying. Most are claiming to have successfully transformed into a big tech or successful unicorn. There is just one problem with that strategy, which is that it isn’t true. Despite massive hiring and spending, with some success of course, they are neither. Their priors embedded in DNA are different, as are inherited strengths and weaknesses. They are simply a different species with different opportunities, risks, advantages, and cultures.
Mastodons are much more skilled at financial engineering than they are in technology, but understand the need to invest to attract talent and provide play boxes to keep people occupied, hence all the talk about innovation. Just don’t expect much change. Mastodons are usually considered strategic to large national economies, are too big to fail, and so may be able to power on by sheer inertia until the next killer asteroid hits earth. Since so few mastodons are left and they typically don’t think they need anyone’s help, I won’t spend additional time on them except to make one point.
The auto industry has invested the most capital in AI to date, and did so earlier than most. Although most of the market leaders started with the mastodon strategy, all but one auto giant has since determined that they weren’t too big to fail, they couldn’t do it alone, and needed help, so they have formed a variety of interesting partnerships. These companies are being forced by technology and disruption to become transformers. The same may or may not be true for other industry leaders. For now, the mastodon strategy appears to be working only for a very few (single digits). Some apparent early success may actually be an effective stalling technique rather than long-term strategy (we should find out within the next five years).
The transformers
I’ve identified three types of transformer companies. The most dynamic and successful are true transformers typically led by a management team and board that well understand the existential risks facing their company. They are rare, but can be found in most industries. These scenarios are massive turn-around efforts, preferably before the companies are facing serious disruption. Transformations don’t work terribly well under severe financial stress. True transformers don’t produce many press releases on the topic as they are focused on the heavy lifting of replacing outdated business models. They depend on results to move share price, not marketing fluff. True transformers tend to make good partners for serious startups due to similar cultures and needs. Those attempting to transform into something different for survival must take greater risks. Management and owners usually have more personal skin in the game. A lot is at stake with true transformers. If they succeed in the transformation, they may become industry leaders. More often than not they don’t succeed and are then forced to rely on investment bankers to find a financial solution in M&A.
Partial transformers
The most common type of transformer is partial, representing the majority of the Fortune 500. Partial transformers enjoy good solid business units that are performing well and are not facing existential risk in the foreseeable future. Think of a leading national restaurant or hotel company—not yet facing the risks of auto, retail, publishing, etc. Partial transformers are a great place to advance one’s careers, particularly as a CXO in mid-career and raising a family. The risks are low and manageable, compensation is good, and one has a reasonably good chance of having a life outside the office and airports. As the name implies, partial transformers do have risk and opportunity, so they are actively experimenting, do some strategic partnering and venture investing, and usually acquire a few companies. Partial transformers usually do well with AI projects with limited goals. They can make great customers due to sufficient financial strength, stability and scale.
Fake transformers
Fake transformers remind me of the old low budget thriller movies where the monsters really don’t look real to anyone. It is almost as if everyone is expected to understand this is just all fantasy theatre, but they hope you buy their stock and products anyway. Fake transformers invest heavily in marketing and PR, buybacks, management compensation packages, and even technology, but underperform their peers and the market. Fake transformers suffer from failed leadership in the CEO and board.
Intelligence agencies
While all companies perform competitive intel, some are more intelligence gatherers than product or service providers. Intel gatherers became quite problematic in AI systems. They are essentially impossible to work with because their intentions are not what they claim. They may act as your best friend to elicit deeper intel, but true intentions typically include your demise. Although this model is obviously not successful as an AI implementation strategy, it can be an effective business model. The company may find an opportunity, partner, or acquirer, or may broker the intel as it may be more valuable than their company. I view this AI strategy as temporary at best because it creates so many enemies.
Vertical maestro
One of the most successful AI strategies to date for incumbent companies has been the vertical maestro. These are usually the market leader in a large specialty industry that is sufficiently technical to require a strong team with consistent investment over decades. Think specialty manufacturing with their own factories and dealership networks. Some of these companies face serious risk from AI combined with other factors, such as emerging competition, political pressure, and significant debt. The tech industry also has many vertical maestros, such as Salesforce. They acquired one of our early peers and have increasingly built their business around AI functionality. Healthcare also has many vertical maestros, though healthcare has been much slower to adopt strategic AI.
These companies were in a good position to leverage machine learning projects and integrate into their existing product lines. However, vertical maestros tend to be optimized in price relative to debt level—sometimes priced higher than the market can withstand. In some cases their customers are seeking alternatives out of necessity. Customers may only have one or two choices when they need a full menu, perhaps even to survive. Therein lies the challenge for vertical maestros—by further consolidating their grip on markets, they may actually be increasing risk of displacement. Vertical maestros make good candidates for lateral moves into other industries, whether through strategic partnerships, investment, acquisition, or creating a new business from scratch.
Platform conglomerate
Amazon is of course the ultimate example of a platform conglomerate, and a fairly effective one at that. However, many platform conglomerates exist, particularly since the commercialization of the Internet, though other forms have existed throughout history. Ecommerce platforms are inherently favorable to conglomeration. Once the network is established, scale is expanding and the company is break-even, expansion into new products and lines of business are fairly easy. Native network companies also have a technical advantage over incumbents due to the lack of legacy IT systems and cultures. Anyone building a new platform today would be foolish not to plan for incorporating machine learning, and would be wise to partner with, or if sufficiently funded, build a complete AI system. Uber is a good example of the latter—they acquired a startup that became their AI lab, which much of the future business was built around, and have since expanded into other areas.
One of the main challenges for platform companies is to resist excessive conglomeration and remain focused on core areas that are profitable and sustainable. I experienced this myself in our lab and incubator in the 1990s when we operated small platforms by comparison to today’s leaders. It was all-too easy to expand into new areas online. Difficult to resist the temptation no matter how wise.
Authentic business builders
These are the most successful new companies in AI now numbering in dozens if not hundreds. Most are specialists such as drug development ventures or autonomous vehicles. A few like my company are focused on universal AI systems. Alexa is a large company example of a universal system even if very different than ours. Tesla is an example of a rapidly growing authentic business builder that integrated AI and became a vertical maestro.
Authentic business builders tend to attract the most driven teams. They are exceptionally mission-oriented and very focused on purpose such as curing disease, preventing crises, space travel, and/or sustainability, whether environmental, economic or as I see it – both. They also tend to be where most of the new wealth is created. I expect future pure play market leaders to emerge from this group. If this doesn’t occur within the next couple of years it is a clear sign of market failure due to the big tech oligopoly, which would cause massive damage to the U.S. economy and future, calling for even more radical antitrust intervention than currently contemplated.
Hybrids
Some of the most successful AI companies are hybrids of the various species. Alphabet for example has an emerging mastodon in Google that is also a platform conglomerate, and owns other companies that are authentic business builders and true transformers. Google is also one of the largest intel gathering networks on the planet. Microsoft, Apple and others have examples in their companies as well that meet the criteria of each of these species, including the occasional fake transformer. Large incumbents should probably employ a hybrid of each type on a selective basis.
Good post and a good discussion on how AI is being used or investigated across the Industry and government…
Thanks Marv. Hope all is well with you and yours.
Thought provoking article on the different entity types and their relationship with AI.