Jack Williams, iTero and GIANTX: When AI coaching becomes an exclusive asset of one esports team

HOOK

In August 2026, at Gamescom in Cologne, Natus Vincere lifted the Aegis of Champions after defeating EHOME in the final of the first The International. Fourteen years later, an esports journalist sat in an office and typed that memory back into his own biography, before beginning an interview about artificial intelligence. The distance between these two events is not just time. It is the distance between an industry that played games for passion and an industry being priced by data.

The interview is titled Jack Williams on iTero, GIANTX and the future of AI coaching in esports. The interviewee is behind iTero, an analytics tool based on machine learning. The partner mentioned is GIANTX, a European esports organisation. The central topic is AI coaching.

What is remarkable is how the interview is structured. The first half discusses working exclusively with GIANTX and the likelihood of being copied. The second half discusses cheating in competition with AI assistance. Not a single performance metric is cited. Not a single percentage of win-rate improvement is disclosed. Not a single comparison between teams that use the tool and teams that do not.

To me, a financial analyst who has spent most of his career scrutinising sports club reports, that silence matters more than any number that could have been offered. In sports, when a technology vendor chooses to talk about contracts rather than competitive results, they are telling you their value lies in the market's structure, not in the product's performance.

Across nearly two decades of watching how sports organisations spend, I have learned a rule: every claim about a "technology breakthrough" in sport can be tested with three questions. How much does the technology improve outcomes? What is the cost of acquiring it? And who is paying? The iTero interview answers the third question clearly. The other two are left blank.

This is precisely where a financial analyst is needed rather than a technology journalist. The story of iTero is not in the algorithm. It is in the power structure of a franchised league, in an exclusive deal with a single organisation, and in the question the entire esports industry is avoiding.

CONTEXT

To understand why this conversation matters, the context must be reconstructed.

iTero, as described in the interview, is an analytics tool equipped with machine-learning models, capable of processing match data and supporting tactical preparation for professional esports teams. It belongs to the product category the industry calls "AI coaching tools". This is not an entirely new market. Top teams in League of Legends, Dota 2 and Counter-Strike 2 have used multiple layers of analytics tools for over a decade, from Riot's data trackers to commercial products such as Mobalytics and to in-house platforms built by large organisations.

What is new is not the technology. What is new is the commercial form.

GIANTX, the organisation named in the interview as iTero's exclusive partner, is a European esports entity with a presence in the League of Legends EMEA ecosystem (LEC). This is a league operating under a franchise model, meaning member teams are not relegated on competitive results but purchase long-term participation slots from the operator. This model has a structural consequence few analysts discuss deeply enough: when a team cannot be relegated, every structural advantage it holds can persist across multiple seasons rather than being competed away by the market.

GIANTX was formed through the merger of two organisations with European history, and its acquisition of an LEC slot was the result of a long commercial negotiation. This is the key point: GIANTX is not a rising team that might dissolve after one failed season. It is a permanent member of a closed league. This makes every advantage it holds, including an exclusive contract with a technology vendor, a governance issue at league level rather than merely team level.

Williams speaks about working exclusively with GIANTX. That is the hinge for a larger question: what happens when a team in a closed league holds exclusive ownership of a tool that may influence competitive outcomes?

Technically, three time windows must be distinguished in which an AI tool can act: pre-match, between games within a BO3 or BO5 series, and in-game as real-time assistance. In every major title today, real-time assistance is explicitly prohibited. The interesting grey zone lies in the between-game window, where a tool can process data from the game just concluded and propose adjustments before the next begins. This is the window coaches call the "fifteen-minute window", and it is where the commercial value of AI tools is actually determined.

The interview references two sections. The first covers working exclusively with GIANTX and the likelihood of being copied. The second covers AI-assisted cheating. These are two sides of the same coin: the commercial side and the integrity side. Both belong to a larger question no league has adequately answered: who owns tactical intellectual property?

One more layer of context is needed regarding market dynamics. Over the past five years, venture funds have injected hundreds of millions of dollars into esports analytics startups. Valuations typically rest on two metrics: the number of client teams and the volume of data processed. Both metrics are inversely related to exclusivity. An exclusive tool serving one team processes less data but creates a larger market advantage for that client. A widely distributed tool processes more data but creates less advantage for each client. This is one of the central paradoxes of the industry, and I will return to it.

Watching technology interviews in sport across many years, I have noticed a recurring pattern. The interviewer asks about the technology. The interviewee answers about the technology. But on closer reading, what truly matters is hidden in the questions that were never asked. The iTero interview is a textbook case. It answers many questions about the product and almost none about the business model.

From the perspective of a club financial analyst, this is a worrying sign. In professional sport, the most important question about any tool is never "how does it work". The most important question is always "what pays for it, and who". The interview does not answer that. But it leaves enough traces for a financial analyst to reconstruct the underlying structure.

CORE

Now to the actual analysis. There are five layers of issues in this interview, and each has a cost-benefit structure that mainstream coverage ignores.

Layer one: what an exclusivity contract actually is, and why it matters

When Jack Williams speaks of "working exclusively with GIANTX", at least three readings are possible. First, GIANTX is the sole client within a specific league. Second, GIANTX is the sole client across an entire title ecosystem. Third, GIANTX has a short-term arrangement to supply feedback and data to iTero during product development.

In B2B software, the third form is the most common and the least discussed. The "design partner" model is familiar in enterprise technology: a startup grants early access and preferential pricing to a first client in exchange for operational data and feedback. In esports, this means a professional team plays the role of beta laboratory. It receives a temporary advantage while the tool remains exclusive, and pays by having its tactical data become part of a product later sold to rivals.

This is the point most esports interviews fail to exploit: the true value of an AI tool is not in what it does for the first team that uses it. It is in what it learns from that first team to sell to the second.

From a financial analyst's vantage, this structure is anything but new. It mirrors how business-data companies operate. Bloomberg, Refinitiv and other financial-data vendors all began by granting free access to a handful of large institutional clients in exchange for their trading data. As the product matured, those clients became paying customers, and their data became part of the value sold to others.

In South Korea, where I live and work, fintech companies followed the same path in the 2010s. Kakao Pay began by offering free payment services to small merchants in exchange for transaction data. Once the merchant network was large enough, it shifted to a fee model. With iTero, the scenario may be similar, except the currency is not won or dollars but tactical data.

So the right question is not "does GIANTX have exclusivity on iTero" but "how much is GIANTX paying, and how". If they receive the tool free or discounted in exchange for data, that is an indirect subsidy equivalent to an investment in tactical infrastructure. Under the financial rules of many esports leagues, such subsidies may not be recorded. A team can "spend" hundreds of thousands of dollars on analytics infrastructure without the figure ever appearing in league reports, because it is paid in data, not money.

I witnessed exactly this structure at Incheon United in 2026. When I proposed a player-valuation model based on social-media data, management rejected it. But only one season later, a Korean sports-analytics company approached the club offering a free tool. The only condition: they could use the data to improve the product. The club declined because it did not understand the value of its data. Had it accepted, Incheon United would have held an analytics advantage costing not a single won on the balance sheet. That is the most dangerous kind of advantage, because it is invisible to every existing financial-monitoring mechanism.

Layer two: copyability and the half-life of advantage

Williams mentions the tool may be copied. That is a sound recognition of the economics of technology advantage in esports: its life cycle is far shorter than in traditional industries.

The reason is not the technology itself. It is patch cadence. This is where structural analysis becomes necessary, because the two largest esports titles operate on completely different rhythms, and that difference determines an AI tool's value.

League of Legends patches every two weeks. Each patch adjusts champion stats, items or game mechanics. This is a cycle so short that any data pattern an AI tool learns last month may lose value after just two patches. In such an environment, the AI's value is not in "solving the meta" but in "detecting the new meta faster than opponents". That is an advantage of speed, not of knowledge.

Dota 2 operates differently. Valve updates major systemic patches far less frequently, but each patch is more volatile. Between patches, there are stable stretches lasting several months. In that environment, a machine-learning model trained on historical data retains validity longer, and the AI's value shifts from "speed" to "depth of historical modelling".

A single AI product marketed identically across both types of titles should be a red flag. Its core value proposition inverts completely depending on the patch cadence of the title it serves.

This is precisely what a company like iTero needs to answer clearly, and the interview gives us no data to assess it. We do not know the patch cadence of the title they target, the training-data time horizon, or the tournament server lock rules of the leagues. Only these three inputs together determine whether an advantage is durable.

In financial analysis, when a company does not disclose these three basic variables, we call it "hidden model risk". The investor does not know what assumptions underpin the valuation. In esports, hidden model risk means teams buy tools without knowing the half-life of the advantage they are paying for. If that half-life is three months, they are paying a long-term asset price for a short-term asset. If it is eighteen months, they have a good deal. Without data, there is no way to distinguish between these scenarios.

This is why I always told the club executives I worked with: Every valuation model is wrong. The question is: wrong in whose favour. When a vendor does not disclose performance data, their valuation model is wrong in their favour. When a team accepts a purchase without asking, their model is wrong against them. This is the basic lesson of every deal in professional sport.

Layer three: the AI-assisted cheating story and the legal grey zone

The second half of the interview concerns AI-assisted cheating. This is the hottest integrity topic, but also the most misunderstood.

In every major title today, real-time assistance during competition is explicitly prohibited. This leaves little room for debate in the in-game cheating frame. The real grey zone lies in the between-game window. In a BO3 or BO5, there is a short interval between two games: enough for a coach or a tool to process data from the previous game and propose adjustments for the next.

How have major leagues regulated this? The answer varies. Some leagues limit the number of staff permitted contact with the team between games. Some prohibit all communication with coaches during the inter-game break. Others allow coaches into the room but forbid devices. All of these rules were designed for the pre-AI world. They were not designed for a world in which a team can prepare an AI model running on a remote server, receive data from an observer in the room, and return analytics to the captain via a smart wearable.

The problem is not whether AI can cheat. The problem is that current rules were written for a world that no longer exists.

In South Korea, where I live and work, StarCraft leagues in the 2010s faced analogous issues in another form: assist software, cheat software and bot variants. Experience from those leagues shows a rule: every ban is defeated by a form not yet defined. Regulation is only effective when it defines the mechanism of violation, not merely the act of violation.

With AI, the challenge is greater. A model can run on a personal computer, a cloud server, or a compact device in the pocket of a fan in the stands. Detecting it requires leagues to invest in digital-monitoring infrastructure, meaning money, personnel and a new rulebook.

On another dimension, the AI-cheating frame touches data privacy. If a league wishes to monitor all digital communication inside the competition area, it must process the personal data of coaches, players and staff. This raises questions under GDPR in Europe and analogous personal-data-protection laws elsewhere. A league cannot simply demand surveillance rights; it needs a legal framework that permits it. This is an aspect esports administrators often overlook, and it may become a major barrier soon.

My experience at Incheon United in 2026, when the pandemic emptied stadiums, taught me something: a crisis does not create new problems; it exposes structures that have long been dead. Esports cheating rules have long been dead, in the sense that they no longer fit technological reality. The AI debate merely makes that clear.

Layer four: the franchise structure and competitive fairness

This is the part I believe is the biggest blind spot in the entire AI debate in esports.

In franchise leagues, member teams are not relegated on results. They purchase long-term slots. This structure produces a consequence I have observed in other sports, especially football leagues without relegation: structural advantages are not competed away by the market.

In an open league, if one team has a technology advantage, rivals can catch up through recruitment, buying tools, or changing their operating model. In a closed league, this dynamic slows considerably. One team can hold an advantage across three or four seasons while the rest of the league has no mandatory mechanism to close the gap.

If iTero grants GIANTX an exclusive advantage during a product's early phase, and if that advantage lasts several seasons, competitive fairness becomes the central question. The LEC operator may have to choose between two paths: force iTero to offer similar terms to every other team in the league, or restrict the tool's use in official matches.

Both paths carry costs. The first distorts the private market of tool vendors, turning them into mandatory suppliers for every team and reducing innovation incentives because every rival gains access to the same technology. The second drives tool use into the shadows, reducing oversight. In other words, no solution is neutral. Every decision redistributes advantage to a group of stakeholders.

In football, FIFA and continental confederations faced the same problem when analytics tools spread in the 2010s. Their approach: rules on devices in the technical area, limits on tablets and communication devices, and mandatory pre-registration of all analytics software with the operator. That is a reference model, not because it is perfect, but because it shows what an operator can do when forced to act.

Esports leagues currently lack an equivalent system. This is a governance gap, not a technical one. At Incheon United, I repeatedly told management we needed a policy framework for analytics tools before such tools became standard. They laughed and said the K-League had no rules on the matter. Two years later, when most clubs had their own data analysts, competitive fairness suddenly became a hot topic in operator meetings. That is the typical governance pattern of professional sport: late adjustment after the problem has taken root.

Layer five: intellectual property and the question of who owns tactics

The least discussed aspect of the entire AI coaching debate is intellectual property.

When a team uses an AI tool to prepare tactics, what data is generated? Who owns it? In most B2B software contracts, operational data generated by the client belongs to the client. But in the AI world, that data is also model food; it makes the model better for all other users. This is the familiar "data flywheel" of software.

For an esports team like GIANTX, this means every tactical analysis it performs through iTero contributes to improving the product iTero will sell to other teams. GIANTX pays to improve a tool its rivals will use. The asymmetry may be offset by first-mover advantage during exclusivity. But once exclusivity ends, long-term value flows to the tool provider, not the team.

This is why the "likelihood of being copied" question in the interview is interesting. It implies the tool provider knows its advantage is not durable. But the question it never asks is: who benefits from that non-durability?

In finance, data companies like Bloomberg and Refinitiv built long-term advantage by locking client data into their products. Fund managers pay hundreds of millions of dollars a year for transaction data, because without it they lose competitiveness. Esports is following the same path on a smaller scale and at a lower level of professionalisation. That is a description of the market's stage of development, and there is nothing surprising about similar business models emerging in similarly structured markets.

Personally, I experienced exactly this asymmetry in the loan deal for Ibrahima Ndiaye during the 2026 Qatar World Cup. Ndiaye's parent club in Ligue 2 undervalued him because they looked only at data from within French football. When we signed a six-month loan with a 60-40 wage split, we held an information advantage the other club lacked. Ndiaye scored seven goals in the second half of the season, saving Incheon United from relegation. The French club realised its mistake only after the season ended. This is the essence of every information market: the creator of data is never the biggest beneficiary of that data.

With iTero, the same structure repeats at a much larger scale. GIANTX supplies data; iTero captures value. In the short term, GIANTX gains a tactical edge. In the long term, iTero holds intellectual property. The two sides are playing different games with the same deck.

This is why leading esports teams in Korea and China have begun building in-house analytics departments instead of buying external tools. They learned the finance industry lesson: if you do not control your data, you are renting rather than owning your advantage. T1, Gen.G and JDG have all invested in in-house analytics teams over the past few years, and that is a sound long-term strategic decision even if it costs more in the short term. Here is one clear example that esports is not football's rival. It is a mirror exposing the entire spending habit of this industry.

CONTRARIAN

Now to the counter-intuitive part. Three popular assumptions about AI coaching in esports are, I believe, wrong, and I will use data from other industries to challenge them.

First counter-intuitive claim: an exclusivity contract is a win for the team. In practice, it is often a long-term loss.

The obvious logic is: a team with exclusivity has an advantage. But that advantage has a price. The team pays with its tactical data, with not having a direct rival to benchmark against in the early phase, and with the risk of dependence on a single vendor. When exclusivity ends, the team still depends on the tool, while rivals may have developed independent processes or diversified suppliers.

This is a lesson from the electric-vehicle industry. Tesla once held exclusivity on superfast charging. When other manufacturers began opening competing stations, that advantage ceased to be durable, yet Tesla's investment had become part of the industry standard. In esports, the first team to adopt an AI tool may be prepaying for infrastructure every other team will later have at a lower price.

In South Korea, I observed a similar dynamic in how K-League clubs invested in GPS player-tracking systems. The first club to buy such a system in 2026 paid three times the current market price. By 2026, every club in the league had a similar system at one third the cost. The first club held no long-term advantage. It was simply the prepayer for infrastructure the whole industry would use.

This means teams must distinguish between two types of technology advantage: advantages defensible by technical barriers, and advantages copyable within eighteen months. For the first, a high price is justified. For the second, a high price is a strategic mistake. Most esports teams cannot tell the two apart, and AI tool vendors have an incentive not to help them.

Second counter-intuitive claim: the AI-cheating debate is a distraction. The real problem is structural inequality.

Esports cheating rules focus on individual behaviour: players using unauthorised software, coaches violating communication bans, fans manipulating data. But the real problem of AI coaching is not individual. It is the difference in access to tools between teams.

A team like GIANTX can invest in AI infrastructure because it has the budget and the network. A smaller team cannot. This difference violates no rule. It is simply a resource gap. But its impact on competitive outcomes can be equivalent to a legitimised form of cheating, because the league permits it.

I saw this dynamic in Korean football at Incheon United. Large K-League clubs have data-analysis departments with dozens of staff. Small clubs have one or two. This difference violates no playing rule but produces an irreparable gap on the pitch. In esports, where the marginal cost of adopting technology is much lower, the gap can be narrowed or widened depending on access policies.

In finance, small fund managers often hold a modest information advantage over large institutions. They cannot compete on data processing speed, but they can compete on focus. In esports, a small team cannot build an enterprise-grade AI model, but it can exploit a small set of deep data a vendor can never have. This is a viable path for small teams, and it is being ignored in the AI debate.

Third counter-intuitive claim: iTero's value is not in AI. It is in data the organisation does not have.

Any team can build its own AI model using open-source libraries and public data. What it lacks is high-quality internal data from rival teams. That is where iTero's true value lies: the ability to aggregate data across many organisations to produce a better model than any single team could build.

This means iTero's business model depends on signing many teams. It also means each new contract reduces the relative advantage of the team that signed earlier. When every LEC team uses iTero, no team has an advantage, and the only remaining value sits with iTero.

This is the paradox of network products. The value of an analytics tool is inversely proportional to the number of teams using it from the team's vantage, and directly proportional to the number using it from the vendor's vantage.

In my first contract as an analyst at Incheon United, I built a valuation model based on social-media data to evaluate the 23-year-old midfielder Kim Do-hyuk. My model showed a 214% growth in Instagram followers over six months, three times that of players with similar professional metrics. Management rejected it as "a fan game". A year later, tracking similar data across multiple clubs, I realised something: the value I had calculated was not in Kim Do-hyuk. It was in the model's applicability to other players. That is the logic of a product, not of a single analysis. And that is also iTero's logic.

I developed three parallel versions of that model. One based purely on social-media data. One combining social data with match data. One based on sponsorship and shirt-sales revenue. All three were wrong in different ways. But placed side by side, I saw a pattern no single version could reveal. That is the most important principle I carried out of that period: the best analysis is not one correct number. It is a set of wrong numbers self-consciously arranged side by side.

With iTero, I suspect they are selling a single version of a problem whose nature demands multiple parallel versions. Because every analytics system is bounded by its assumptions, and every machine-learning model is bounded by its training data. A tool can solve a specific problem very well but cannot automatically detect that the problem has changed. That is human work, and it is why AI coaching will never replace coaches. It can only expand their capability or make them dependent on a tool they do not understand.

Over years of tracking financial deals in sport, I have seen this repeat. A tool is offered as a total solution. Teams buy out of fear of being left behind. Six months later, they realise the tool cannot handle a specific situation that matters to them. But the contract is signed, the data has flowed to the vendor's servers, and they cannot return. This is the nature of every technology market in professional sport. It is nothing novel. It arrived in esports ten years later than in other industries.

TAKEAWAY

The final question is not whether AI coaching will become part of professional esports. It already is. The question is who will control that infrastructure, and how much leagues will pay for not controlling it.

In football, national and continental federations took over a decade to adjust rules for the data era. In esports, where operating processes are more flexible, the adjustment could be faster, or slower if stakeholders choose to ignore the issue until a large incident forces action. The Jack Williams interview poses the question without answering it. It shows that even tool builders recognise their advantage is temporary and that copying is inevitable.

What the interview does not say is: when there is no exclusive advantage, what will teams compete on? The answer, based on two decades of watching professional sport, is things that cannot be digitised. Team culture, psychological resilience, decision-making under pressure. Factors no machine-learning model can teach, yet every machine-learning model can crush if it is not trained to respect them.

Looking further ahead, there is a question I have not seen anyone pose in the AI-coaching debate. If the value of an AI tool depends on teams' tactical data, and that data is increasingly harvested from the teams least able to protect it, then in ten years, who will own the tactical map of an entire league? This is a question the esports industry is not ready to answer, and may never answer until the answer becomes too obvious to ignore.

That means the next game is not a game of technology. It is a game of those who know how to use technology without letting technology define them. And that is a game no tool vendor can sell you.

I am still watching. I am still here in Incheon, red pencil on the financial report, questioning every number, every cash flow, every contract hidden behind a beautiful headline. And I still believe that in every sports market, from K-League to LEC, from football to Dota 2, the winner is not the biggest spender. The winner is the one who understands best where every dollar is going.