Taiwan’s Bicycle Industry Has a Cognition Problem

A framework from AI research puts the industry’s structural weaknesses in sharp diagnostic terms — and points to what needs to be built.

The last few posts have focused on the nature of change and what the bike industry needs to take into account in responding.

Here’s one more in the series in which we examine the bike industry through the lens of recently published framework and blueprint to the future.

Note that these are speculative and exploratory in nature. What the shape of the business future is (for the bike industry and everyone else) is unknown. What is guaranteed is massive transformation is underway.

Three Claims

In 2025, Dr. Alexander Wissner-Gross and Dr. Peter Diamandis published Solve Everything: Achieving Abundance by 2035, a blueprint for how AI will industrialize entire domains of human activity over the next decade. 

The core argument is not primarily about technology. It is about cognition: who controls it, how it gets measured, and what happens to industries that fail to organize it systematically.

It’s also a lens through which to evaluate and manage change.

The framework rests on three claims.

The first is that cognition is becoming a commodity. Expert attention — the scarcest input in any complex supply chain — is being repriced downward by AI. 

Industries that treat decision-making as an artisanal skill, relying on experienced humans with good instincts, are approaching a structural disadvantage against industries that have turned decision-making into a measurable, repeatable process.

The second claim is that progress industrialises only when it has targeting systems. Not goals. Not strategies. Targeting systems: rigorous, adversarial benchmarks that define success with mathematical precision, attach financial consequences to outcomes, and operate continuously rather than as annual reviews. 

The protein-folding problem in biology was stuck for fifty years. Once the CASP competition gave the field a public, blinded, adversarial benchmark, DeepMind poured scaled compute into the stack and the domain collapsed. What looked like a miracle was a predictable engineering outcome.

The third claim is about direction. 

Raw AI capability is like explosive energy, powerful but unfocused. 

The framework calls for “shaped charges”: channelling that capability through specific, measurable missions rather than letting it dissipate into marginal efficiency gains.

The mechanism connecting these three claims is a maturation curve running from L0 to L5.

L0 is the “Muddle”. Goals are ambiguous. Data is messy or absent. Decisions are driven by instinct, relationships, and whoever makes the most convincing argument in the room. Every success feels like a lucky accident.

L1 is the scoreboard. The industry has agreed on what to measure, even if it does not yet know how to improve it systematically. Export numbers. Market share. Quarterly turnover. The numbers exist. The shared infrastructure to act on them does not.

L2 is standard operating procedures. The best performers have identified patterns, have written them down, and made them repeatable. The process is still manual, but it is consistent. Variance is reduced.

L3 is automation. The SOPs from L2 become code. AI handles the routine work. Humans move up the stack, focusing on exceptions and strategy rather than execution.

L4 is the economic flip. Buyers stop purchasing effort and start purchasing verified outcomes. Contracts are tied to results, not hours worked or units shipped.

L5 is commoditised. The problem is compute-bound. Multiple providers can deliver it reliably and compete purely on price. It runs in the background like tap water.

The framework’s central warning: industries that stay in the Muddle do not simply stagnate. Cognition gets commoditized around them by actors who move up the curve faster. The strategic window is not permanent.

The Bicycle Industry 

Applied to Taiwan’s bicycle sector, the framework is unsparing.

The Taiwan Bicycle Association publishes export data. Shimano publishes financial results. Individual companies track turnover. That is scoreboard territory, or in terms of the structure put forward by the good doctors, L1. 

But shared, real-time data flowing between the parts of the supply chain that would need it most is almost entirely absent. 

Retailer sell-through data rarely flows directly to Taiwan OEMs, and brands often share forecasts rather than full inventory visibility with suppliers. 

As a result, when demand shifts, each tier of the bicycle supply chain reacts independently, amplifying the signal through a classic bullwhip effect. And the industry has been living with the consequences since 2022.

When Covid demand surged, the industry amplified those signals at every step, ordered far too much, and by the time reality caught up, warehouses from Taichung to Rotterdam were full. By late 2022 the bust was in full swing. 

Three years later, the industry is still climbing out. At Taipei Cycle 2026 in March, the Show Daily described the mood among 900-plus exhibitors as “cautious resilience.” 

That is a polite description of an industry still in survival mode, now dealing with Iran-related logistics disruptions and US tariff uncertainty stacked on top of an inventory hangover that never fully cleared.

Schwalbe’s Chief Sales Officer Nico Simons put the core problem clearly in a August 2025 industry survey. The most pressing challenge was “realigning supply chains with sustainable, realistic demand levels, and doing so through industry-wide coordination to avoid the resurgence of bullwhip effects.”

That is a precise description of an L0 problem. The industry still has no shared targeting system that would let it distinguish real demand from perceived demand. The data to build one exists in fragments across hundreds of companies. The will to pool it has not yet materialized.

Most of Taiwan’s bicycle sector sits between L0 and L2, depending on the company and the function. 

The manufacturing processes at the better factories are closer to L2 — there are procedures, they are applied, quality is tracked. The supply chain as a whole is closer to L0. 

The governance layer, where labour practices, carbon accounting, and compliance sit, is largely pre-L1: the metrics are not yet agreed on, let alone shared.

The Tariff Problem Is Also a Data Problem

The tariff situation illustrates the same gap from a different angle.

When the Trump administration began imposing reciprocal tariffs in 2025, Taiwan’s bicycle industry entered what Tern’s Josh Hon has described as “limbo.” Not because the tariffs were necessarily devastating — for most Taiwan companies, direct US exposure is limited — but because nobody could calculate the actual impact. 

Merida’s senior vice president Daryl Chang noted his company’s US exposure was modest. Velo Enterprise CEO Ann Chen said the situation was outside the industry’s control. SRAM Asia’s Bob Chen said his company needed more time to evaluate.

Three major players taking three different positions, but none of them grounded in a shared model of what was actually happening.

This is the Muddle in real time. A supply chain operating at L2 or above would have financial models capable of calculating tariff impact within a defined range within days of a policy change. The scenario-planning tools exist. The industry is not using them systematically.

The governance layer matters here too. 

Moving up the maturation curve requires what the framework calls outcome-based contracts; a shift from paying for effort to paying for verified results. 

The OEM and ODM relationships that underpin Taiwan’s supply chain are almost entirely effort-based. A factory gets paid per unit. There is no commercial mechanism to reward a supplier for zero-defect throughput, verified on-time delivery, or audited carbon reduction. 

The EU’s forthcoming carbon border tariff will force some version of this — but the industry is not building the infrastructure proactively.

The Governance Problem: Hiding in Plain Sight

The Giant forced-labour case from September 2025 is worth examining through this lens.

US Customs and Border Protection issued a Withhold Release Order against Giant’s Taiwan-manufactured products, citing allegations of forced labour involving migrant workers. Giant denied the allegations firmly, pointed to its zero-recruitment-fee policy implemented in January 2025, and announced third-party audits. (Giant continues to dispute the allegations and is attempting to demonstrate compliance and remediation.)

What the case revealed was not primarily a labour problem — Giant’s own account suggests it had already addressed the specific issues raised. 

The problem was a verification problem. The industry lacked independent, real-time auditing infrastructure that would have made CBP’s concerns either confirmable or dismissible within days rather than months. The absence of that infrastructure is what made the situation both legally and commercially damaging.

In framework terms, this is the cost of operating without Decision Records, immutable, auditable logs explaining exactly what decisions were made, when, and on what basis. 

Giant could not prove what it claimed to be true about its own practices, not because the practices were wrong, but because the documentation infrastructure to verify them did not exist.

VP Components, a mid-sized Taiwanese component manufacturer, offers a partial counter-example. VP has implemented zero-replacement fees for foreign workers, conducts vendor audits of its tier-two suppliers, and has built recyclable and low-carbon materials into its manufacturing processes. 

It is operating closer to what the framework would call an L2 governance model: there are procedures, they are applied, they are documented. If VP can do it, the argument that it is structurally impossible for the broader industry weakens considerably.

The Data Platform That Already Exists

The most direct example of the maturation curve in practice is Bicycle Cluster, launched by Alfred Tsai out of frustration with how inefficient component sourcing was across the Taiwan supply chain.

In late 2024 it unveiled an Industry Data Platform at the Taiwan Bicycle Association’s general meeting. the IDP is a structured database integrating with the Taichung Bike Week website and supplier sites, linking the supply chain through one unified system. 

In 2025 it added AI-powered chatbots trained on industry-specific data, allowing suppliers to manage product information once and have it become usable across their customer-facing channels.

Tsai is explicit about what he is trying to do. “Our mission from the start was to make data useful in an industry where it was often overlooked. Now we’re proving that structured data — when integrated into real workflows — can drive transformation across sourcing, marketing, and customer support.”

In framework terms, Bicycle Cluster is building the Task Taxonomy layer — breaking complex sourcing jobs into measurable, searchable, automatable actions — and coupling it with an Actuation layer that gives the AI practical tools to affect real purchasing decisions. 

It is not yet an industry-wide targeting system. But it is the most concrete proof of concept that the infrastructure can be built, by a small team, inside the existing Taiwan supply chain ecosystem.

The Show Daily’s year-ahead analysis for 2026 made a related point about AI more broadly: the most valuable near-term applications would not be consumer-facing features but invisible efficiency — forecasting, demand planning, production optimization, automated quality checks. 

That is exactly the L3 transition the framework describes. AI handling the grunt work. Humans focusing on strategy.

What the Framework Says the Industry Needs to Do

Mapped against the maturation curve, the priorities are not complicated. They are just hard.

Firstly

The first and most urgent is establishing shared, real-time demand signals across the supply chain. 

The bullwhip has cost the industry billions over the past three years. 

It is fundamentally an information problem: each link in the chain is making decisions based on noisy, delayed signals from the next link. Fixing it requires pooled point-of-sale data, agreed metrics, and a neutral governing body to manage access. 

The Taiwan Bicycle Association is the obvious candidate to anchor this. The BAS, with 80 members already coordinating on sustainability targets, has demonstrated that collective agreement on shared goals is achievable even among direct competitors. 

The same model applied to demand data would be worth more than any individual product innovation in the current environment.

Secondly

Building verifiable compliance infrastructure before it is legally required. 

Giant’s WRO, the EU carbon border tariff timeline, and ongoing investigations into labour practices at Taiwan companies, are three separate pressure points converging on the same vulnerability. 

The industry does not have the audit infrastructure to prove what it already claims about its own practices. 

Building that infrastructure now is cheaper than the reputational and legal exposure of building it reactively.

Third

The third is the shift to outcome-based commercial models, starting with ESG. The EU carbon tariff deadline gives this a fixed timeline. 

BAS members have committed to reducing CO2 per bicycle by 25 kilograms by 2040. The next step is attaching verified financial consequences to those commitments — funding released when benchmarks are independently confirmed, contracts tied to certified carbon performance. 

The framework calls this Compute Escrow logic: pre-committed resources unlocked only when a specific target is cleared. 

Applied to sustainability, it converts a marketing initiative into a commercial architecture.

What the Framework Cannot Fix

The Wissner-Gross model assumes actors large enough to invest in targeting systems and governance infrastructure. Most of Taiwan’s supply chain is not built that way.

A hub manufacturer in Yuanli or a saddle maker in Changhua — 30 to 200 employees, family-owned, operating on thin margins — does not have the capacity to build its own data infrastructure. 

The fixed costs of targeting systems, audit rails, and demand-signal platforms need to be carried at the industry level and offered to SMEs as a subscription service rather than a capital investment. Bicycle Cluster is already moving in that direction. TBA and BAS could accelerate it.

The framework also assumes that commercial logic is the primary driver of decisions. In Taiwan’s bicycle industry, the primary driver is relationships — long-term OEM partnerships, trust built over decades, the social fabric of how business actually gets done in Taichung. Outcome-based contracts and third-party audits can feel threatening to that model. Reform that ignores this will fail regardless of its technical merit.

The practical approach is to run the new infrastructure alongside existing relationships rather than positioning it as a replacement. 

Use it to verify what companies already believe to be true about their own performance. 

That framing changes the proposition from interrogation to confirmation — and confirmation is something most companies will pay for.

The Bottom Line

Taiwan’s bicycle industry is excellent at making components. The manufacturing base, the component density, the accumulated know-how in Taichung are genuine structural advantages that have not eroded.

What the industry is not good at is measuring what it does, sharing what it knows, and building the commercial infrastructure to prove its own claims.

The Wissner-Gross framework is clear about what happens to industries that stay in the Muddle. 

Cognition gets commoditized around them. 

Companies that move up the curve — by establishing shared benchmarks, building verifiable compliance systems, and shifting from effort-based to outcome-based contracts — create defensible positions. 

Those that wait for the next demand cycle to rescue them are making a bet that the next crisis will be smaller than the last one.

At Taipei Cycle 2026, the conversation was already shifting from volume to value, from hardware to concepts, from survival to strategy. That shift is real. The infrastructure to support it is not yet built.

Building it is the actual work. And crucial work at that!