Food Delivery Platforms: One Tap to Your Door — Who Bears the City's Rush?

On May 1, 2021, delivery workers marched on Ketagalan Boulevard alongside the labor contingent. At that same moment, meals in Taipei were still leaving restaurants, crossing roads, and arriving at the doors of people who could not or did not want to go out. That year, the number of food delivery workers in Taiwan had neared 88,000. ¹

A few years earlier, ordering a meal was like entrusting daily life to a neighborhood shop's phone line. By the pandemic period, the share of food-and-beverage businesses offering delivery and takeout rose from 43.3% in April 2019 to 67.0% during the Level 3 alert in June 2021. Even as dine-in gradually resumed in May 2022, 64.6% of F&B businesses still provided delivery service. ²

Caption: The Fair Trade Commission resolved to prohibit Grab's acquisition of foodpanda's Taiwan business on October 21, 2026; the originally planned second-half 2026 transaction and 2027 integration timeline no longer stand. Image and event information are based on Central News Agency reporting. ³⁴

Delivery is therefore not merely "someone fetches your food." It connects the restaurant kitchen, the platform's algorithm, the delivery worker's phone and scooter, the city's roads, and the recipient's doorstep into a single line. That line shortens the wait, but also concentrates costs that were once scattered across different places onto one invisible person.

📝 Curator's note: The most important thing about delivery is not how convenient it is, but who actually fulfills that "convenience."

A Bunk Bed in a Youth Hostel, Catching a Delivery Income

In 2021, the Social Innovation Platform interviewed a young man using the pseudonym "A-hsiang." He did odd jobs at a restaurant by day, then rode a bicycle delivering food by night. His home was not a walled studio apartment, but an upper bunk about 300 centimeters off the floor in a youth hostel in downtown Taipei. The bunk cost roughly NT$6,000–8,000 per month, required no rental deposit, and sat near the city center where he picked up orders. ⁵

This scene translates "flexible work" into something tangible: a bunk, a bicycle, a phone, and two-plus hours staying online after his day job ended. For A-hsiang, delivery income had once meant repaying student loans, saving money, even the possibility of moving off that upper bunk. After the platform changed its pay calculation, he faced idle waiting, fewer orders, and shrinking earnings — and the vaunted flexibility narrowed with them. ⁵

One case cannot represent all delivery workers, but it reminds us that "freedom" on the platform has never been suspended in midair. It hangs on rent, student loans, transport, and the next order. When any one of those links shifts, so-called freedom can turn into having no choice but to keep accepting orders.

A Single Meal Actually Crosses an Entire City

A 2023 study on Taipei delivery workers interviewed workers, merchants, and platform staff, with the researcher also participating in delivery labor. The study describes delivery as an assemblage jointly composed of digital platforms, phones, scooters, delivery equipment, roads, buildings, users, and merchants. ⁶

This observation matters because delivery workers do not work only "inside the platform." They must locate the shop entrance, judge whether a building allows scooter parking, identify addresses and floors, and negotiate the friction among elevators, rain, scooters, and insulated boxes. Gaps the system has not pre-handled are constantly patched by the person in motion. The study therefore argues that piece-rate pay and the acceleration structure of the scooter city may push more burden and risk back onto the delivery worker. ⁶

Merchants are likewise embedded in this assemblage. The Ministry of Economic Affairs' Department of Statistics notes that the pandemic and changing consumption habits accelerated F&B adoption of social media, mobile payment, and online ordering. Delivery has shifted from a stopgap measure to an operational mode for expanding sales reach. ²

📝 Curator's note: The platform did not replace the corner restaurant; it rewired the corner restaurant into every person's living room.

These figures do not mean everyone orders delivery every day, nor can they directly derive platform market share. They only establish one narrower but more reliable fact: for the F&B industry, delivery to the door is no longer an occasional add-on service.

Taipei's Roads Saw the Risk First

Delivery brings meals to the door, but also brings work onto the roads. Cross-agency inspection data released by the Directorate General of Highways (DGH), Ministry of Transportation and Communications, shows that from January to September 2021, Taipei City recorded 1,567 scooter accidents involving delivery platforms, with 1,212 dead or injured. Over the same period, police cited 7,113 delivery-platform violations. These numbers cover only a specific area and period and cannot be extrapolated to full-year nationwide statistics, but they suffice to show that road safety is not a problem solvable by the delivery worker's individual caution alone. ⁷

The DGH noted in the same press release that since delivery-platform scooters use roads for transport, operators should bear supervisory responsibility. Follow-up measures included physical safe-driving training for scooter riders and mandatory retraining for delivery workers involved in accidents. The key pivot here: delivery is no longer treated merely as a transaction between platform and consumer, but is placed back into the context of public roads and public safety. ⁷

A 2021 reader submission to The Reporter compiled another set of workplace data: the author cited a 2020 delivery worker survey showing 24% of respondents worked 50+ hours per week delivering, and 81.7% worried about traffic accidents. The article also noted that actual coverage of safety education, protective equipment, and heat/cold hazard prevention was inconsistent. This is the advocate's compilation and judgment, to be read separately from official statistics, but it gives shape to the work pressure behind "hurry up a little." ¹

The Word "Subcontracting" Cannot Contain the Whole Road

A 2019 article from the Taiwan Labor Front records that the Occupational Safety and Health Administration (OSHA), Ministry of Labor, launched special labor inspections targeting foodpanda and Uber Eats, and determined that an employment relationship existed between delivery workers and the platforms. The platforms contended the relationship was subcontracting or mere matchmaking. The article also catalogued how uniforms, standardized insulated boxes, scheduling, and inability to refuse orders became flashpoints in judging command-supervision and organizational subordination. ⁸

This is not a mere semantic dispute. If a person can ostensibly go online at will, but in practice is influenced by acceptance-rate metrics, scheduling, pay algorithms, service ratings, and deactivation mechanisms, then "being your own boss" and "having work controlled by the platform" may coexist. The 2016 Constitutional Court Interpretation No. 740 on labor contract determination also emphasized that one cannot look only at the contract title, but must examine the mode of labor provision and business risk. ⁹

The Taipei study archived in Airiti puts this contradiction in street-level terms: delivery workers indeed need on-the-spot judgment, movement, and gap-filling, and thus possess agency. But that agency can also be absorbed by piece-rate pay and the accelerating city, ultimately turning into system costs borne by the individual. ⁶

📝 Curator's note: The real question is not whether delivery workers have freedom, but who must pay when that freedom fails.

Grab Acquires foodpanda — A Different Company Behind the Button

In March 2026, Grab announced it would acquire foodpanda's Taiwan business for approximately US$600 million, about NT$19.3 billion. On March 27, Grab submitted an application to the Fair Trade Commission (FTC). The FTC formally accepted the case in mid-June, stating review would focus on post-merger market impact, monopoly concerns, and whether new competitors could enter the market. ³

This transaction touches daily life not just because two brand logos might swap places. The partner restaurants, users, delivery workers, and order data held by the platform would consolidate with the platform merger. Merchants might face new commission and exposure rules; delivery workers might face new dispatch and pay systems; consumers might re-choose in a more concentrated market. The latter three are life impacts requiring scrutiny, not outcomes that have already occurred — speculation must not be written as fact.

On July 22, 2026, the FTC formally announced a 60-working-day extension of the review period to October 27. The FTC noted the post-merger market structure would remain the same "duopoly" as before. Meanwhile, Uber holds roughly 13% of Grab's shares and 3.7% of voting rights; the FTC needed further analysis of whether this equity link affects Uber Eats' and Grab's incentives and ability to compete, and could create unilateral or coordinated effects. ¹⁰¹¹

Grab told Central News Agency that if approved, the transaction was expected to close in the second half of 2026, with platform integration completed in the first half of 2027, migrating all original foodpanda users to the Grab app. This is the company's proposed plan, not a timeline approved by the FTC. While review was ongoing, consumers could not treat "will definitely migrate" as a foregone conclusion. ¹² This plan ultimately did not occur; the FTC resolved to prohibit the merger on October 21. ⁴

The same CNA report also summarized Grab's statements on governance and equity: Grab is headquartered in Singapore and listed on the US Nasdaq. Uber holds about 13% of common shares but voting rights below 4%; Uber does not participate in Grab's daily operations, and Uber-nominated Grab directors recuse from all board decisions related to Taiwan. These are Grab's public explanations; the FTC must still independently judge based on substantive control and competitive effects. ¹²

Another key to the transaction is not "whose app color changes," but how data, partner merchants, delivery workers, and consumer relationships are rearranged. CNA reporting notes that Grab and foodpanda's integration plan involves migrating original foodpanda users to Grab. As for how commissions, dispatch, pay, and merchant contracts might change, existing sources provide no confirmed proposals; this article therefore does not write potential impacts as established facts. ¹²

Caption: Transaction schematic photo attached to CNA's March 30, 2026 report, directly corresponding to Grab's application to acquire foodpanda's Taiwan business. ¹²

This case must also not be conflated with the 2024 case where Uber Eats directly sought to acquire foodpanda. In a 2024 public article, the FTC stated that the anti-competitive harms of the Uber Eats–foodpanda merger outweighed overall economic benefits, and therefore prohibited that merger. The Grab case was still under review as of August 15, 2026; on October 21 the FTC issued a prohibition decision, but the reasoning differed from 2024 — this time the key was the equity link between Grab and Uber, not the simple addition of two delivery platforms' market shares in Taiwan. ¹³⁴

📝 Curator's note: "Choice" in the delivery market appears not only on the consumer's phone screen, but also in whether platforms still have enough competitive distance from each other.

October 21 — The FTC Says No

On October 21, 2026, the FTC Commission meeting deliberated the Grab–Foodpanda (富胖達) merger and voted: three commissioners favored prohibition, one favored conditional approval with remedies, and one recused. The merger was prohibited on four grounds, announced that afternoon by spokesperson Hsin Chih-chung. ¹⁴⁴

The FTC's core concern was not "what market share Grab would have in Taiwan after buying foodpanda." It noted that Uber, parent of Uber Eats, holds about 13.1% of Grab; Uber's former CEO served long-term as a Grab director; the two have close interaction and commercial cooperation in other markets. Although Grab uses a dual-class share structure and Uber's voting rights are only about 3.9%, this equity link would alter the two companies' interaction strategies, aligning their interests. ⁴

The four grounds: reducing Uber Eats' incentive to compete aggressively; reducing Grab's incentive to compete aggressively; past conduct and interaction patterns increasing coordination risk; lack of countervailing power to disrupt sustained coordination. The FTC specifically cited the 2018 case where Grab acquired Uber's Southeast Asia business, noting both parties proceeded with a difficult-to-reverse transaction despite knowing or should have known it might violate competition law. ⁴

The FTC also evaluated the "conditional approval" path and discussed remedies with the applicants multiple times. Applicants offered share dilution, restricting Uber personnel from serving as Grab directors, and restricting exchange of competitively sensitive information. The FTC found these measures could not ensure effective execution and could not eliminate concerns; the FTC's position was that Uber's stake in Grab must be further reduced, even fully divested, to potentially lower coordination risk. ⁴

The final resolution prohibited the merger under Article 13, Paragraph 1 of the Fair Trade Act, on grounds that anti-competitive disadvantages outweighed overall economic benefits, and the proposed remedies were insufficient to ensure overall economic benefits exceeded disadvantages. ⁴

Grab responded in writing the same day, expressing "disappointment with this outcome," disappointed at being unable to bring innovative services, fair income opportunities, and optimized service standards to Taiwan's consumers, delivery workers, and merchants as planned; and emphasized its Southeast Asia core business remains strong, and it will continue investing in its established growth blueprint. ¹⁵ Taiwan's delivery market thus retains its original duopoly structure, with foodpanda continuing operations under its original brand.

Source: Footnotes 10, 11, 14, 15, 16 in this article

After Platforms Became Daily Life, Institutions Began Catching Up

In June 2026, the government portal My E-Government compiled labor rights for delivery workers and freelancers, placing labor/health insurance, occupational injury insurance, and the food-delivery-platform worker rights zone in the same service context. This compilation itself reveals institutional change: delivery workers are no longer vague identities in the startup sector, but workers who need to be placed into social insurance, occupational injury, and labor protection systems. ¹⁶

But writing institutions into existence does not mean risks have vanished. The Reporter notes that since 2019, the government has successively issued food-delivery operational safety guidelines covering traffic accident prevention, heat hazards, dispatch, and insurance. The same article also uses survey data to question the gap between actual education/protection and the rules. Regulations, enforcement, and the streets delivery workers face every day still need to be examined together. ¹

A public compilation of Legislative Yuan issues also shows that delivery-platform governance involves not only labor rights but also market competition, restaurants, consumers, and platform merger impacts. The issue page records legislators' questions on market concentration and cross-border platform influence, and proposals for a dedicated delivery-service law. These are public positions of political figures, not yet established judicial or economic conclusions. ¹⁷

Delivery platforms have thus become a very particular form of living infrastructure: when we use them, we only see the meal status change from "preparing" to "delivered." But between those two states lie the restaurant's plating speed, the platform's dispatch logic, the delivery worker's route, road safety, and whether institutions truly let a person get protection after an accident.

Three Kinds of Waiting the Phone Screen Doesn't Show

Consumers see a simple progress bar: merchant accepts, food prepares, rider picks up, arriving soon. But the Airiti study reminds us these statuses are not a single flow, but coordination among many objects and people. An address on the phone must be translated into the shop doorway, elevator, stairs, security guard, and the recipient's actual location. ⁶

Merchants face a different wait. The Department of Statistics lists delivery business share alongside online ordering, mobile payment, and social-media management, showing the F&B industry faces not a single "whether to join a platform" choice, but how to connect kitchen, orders, payment, and delivery. ²

Delivery workers face a third wait: wait for platform dispatch, wait for merchant to plate, wait for elevator, wait for a safe gap in traffic. These times are not all visible in a single order, yet they affect how many orders a day can be completed, and affect income and fatigue. In the profile, A-hsiang's idle waiting and few orders are precisely this interface-hidden time suddenly surfacing. ⁵

Therefore, delivery-platform efficiency cannot be measured only by "minutes from order to delivery." Research, statistics, and profiles read together reveal that part of that efficiency is actually completed by delivery workers patching gaps on the spot. ²⁵⁶

The City Begins Taking Responsibility for the Button

The DGH's inspection placed delivery platforms into road management, rather than blaming accidents solely on the rider. The government portal placed delivery workers into the labor/health insurance and occupational injury insurance service context. Two institutional languages — one handling roads, one handling work — together come closer to delivery's real position. ⁷¹⁶

The Legislative Yuan issue compilation's presentation of merger and market-concentration discussion also shows the platform is not a neutral screen. When platform scale, algorithms, and the dependency of restaurants, consumers, and delivery workers all grow together, competition policy and labor policy meet inside a single life service. These political claims must still be separated from formal regulations and competent-authority decisions, but they point to the direction public debate is pursuing. ¹⁷

A 2019 Taiwan Labor Front article condensed the problem into an institutional puzzle: workers appear to gain scheduling freedom, yet may simultaneously lose employer responsibility and occupational injury protection. This question remains worth keeping today, because every time a platform changes dispatch, pay, or rating methods, it rearranges who bears uncertainty. ⁸

Before the Next Tap on "Place Order"

Taiwan's delivery life need not be written as "technology makes everything better," nor reduced to "platforms only exploit." The Ministry of Economic Affairs' statistics confirm the F&B industry indeed expanded service through delivery. Research and profiles confirm this convenience must pass through concrete people, vehicles, roads, and homes to arrive. Both sides are true; placed together they approach the full picture of life. ²⁵⁶

This also explains why a good delivery article cannot cite only the platform's own reports. Government statistics answer what share of F&B businesses offer delivery; traffic agencies answer accidents and enforcement; academic research dissects the labor assemblage; profiles bring institutional numbers back to one person's bunk, bicycle, and night. They cannot replace each other, but they can mutually constrain oversimplified conclusions. ²⁵⁶⁷

So, the next time you tap "Place Order," what's worth remembering is not whether to order delivery, but where that convenience comes from. When a city compresses waiting into minutes, it should also have the capacity to put fair pay, safe roads, and accountable responsibility back on the same road.

This division of labor also reminds readers not to write "consumers, merchants, platforms, delivery workers" as four unrelated roles. Research interviews already placed delivery workers, merchants, and platform staff in the same field site. The DGH's inspection also placed operator responsibility and road use together. ⁶⁷

When institutions manage responsibilities separately but life still binds them into a single order, problems recur at the boundaries. Seeing the boundaries is the starting point for understanding Taiwan's delivery daily life. ⁶⁸

This is also why institutions must keep updating, not just patch holes after accidents. ⁷¹⁶

In that 2021 labor-march photo, delivery workers were not abstract "platform manpower." They wore their own clothes, stood on their own roads, demanding this city acknowledge: what arrives at the door is not just a meal, but the time a person left behind crossing the city.

References