How long does it take to make a T-shirt? The answer depends on which clock you start.

At the sewing line, a conventional cotton T-shirt requires eight person-minutes of direct labor across its sewing operations, according to Michael Hofmannrichter, co-founder and CEO of textile robotics startup Silana. This figure captures the total human input across a production line, depending on the number of people involved.

While fashion has spent decades trying to shave seconds off sewing time, this occupies only a fraction of a much longer timeline for getting a shirt to the shop floor. Designing a garment, sourcing the fabric, approving samples, placing an order and moving it to market can unfold over six to nine months.

And now, the industry has another race to contend with: how long it takes to get a garment back on shelves after it’s been returned.

On both sides of the coin — in linear and circular supply chains — time is money, but the relationship between speed and sustainability is not that simple. Faster decisions and more flexible production can let brands commit to holding less stock, replenish the hits that sell, and avoid some of the markdowns and unsold inventory that are routinely baked into seasonal forecasting.

But often, the more efficient something becomes, the more we do, meaning the same infrastructure that is making fashion production faster can also lead to more styles, more frequent drops and greater total output — unraveling the sustainability case for faster production.

The wrong clock

In apparel sourcing, cost per minute doesn’t actually refer to the number of garments produced every 60 seconds. It’s the amount a factory charges for a minute of production, including overhead costs and profit.

To calculate the manufacturing cost of a garment, the factory multiples the cost per minute by the standard amount of minutes it takes to make the garment. Neither measure captures the cost of waiting for a design decision, a fabric commitment or an order confirmation — or the financial hit when the garment fails to sell.

“Fashion’s conundrum is risk, meaning the high cost of being wrong,” says John Thorbeck, founder and chairman of supply chain advisory firm Change Capital. The two metrics that increasingly separate winners, he argues, are forecast accuracy and working capital, and any gains are often unevenly distributed across stakeholders. A brand may accept a higher unit price and still improve its margin by buying fewer garments, selling more at full price, and holding less inventory.

For the factory, smaller runs mean more frequent setup and less time producing garments; unless the price covers that work or successful styles lead to larger repeat orders, the brand’s flexibility comes at the supplier’s expense.

The biggest potential for time saving isn’t the sewing time, but the buying process, says Thorbeck. Some brands are expanding their use of advance material and capacity commitments, while delaying final decisions on color, SKU quantities or market allocation. The approach gives suppliers greater visibility while allowing brands to respond to fresher demand signals, potentially compressing the time from design to market by months.

Shein’s Large-scale Automated Test and Reorder model pushes that principle to an extreme. New products are introduced in batches of roughly 100 to 200 units, according to the company, before real-time demand data determines which styles are replenished, sometimes within days.

The advantage of this system is that it reduces waiting time across design, ordering, production and replenishment, rather than pressuring workers to sew faster, explains Dr. Sheng Lu, professor and director of graduate studies in fashion and apparel studies at the University of Delaware.

In 2025, retailer Asos reported that its own Test & React program, which can take a product from design to site in as little as three weeks, accounted for more than 20% of own brand sales. The company reduced stock to roughly £400 million, around 60% below its 2022 peak, while adjusted gross margin rose 370 basis points year-on-year.

Asos attributes that improvement to a wider turnaround that included higher full-price sales, fewer markdowns and lower inventory, though it does not isolate Test & React as the sole cause.

But producing in smaller batches doesn’t necessarily lower costs, because the development and set-up costs are spread across fewer units. “A small batch is usually on the line for half a day at most, so that upstream work is where the real cost and time now sit,” says Shruti Grover, co-founder and CEO of production-planning platform MannyAI.

Clarifying the technical pack, preparing the lay plan, sampling, costing, and ordering fabric can consume more resources than the production run itself.

For the economics to work, brands and factories need agreements that balance small experimental orders with larger repeat business, and share the risk of test runs that don’t gain enough consumer traction to scale. This is where ethical purchasing practices become critical.

Can machines make production faster?

Machine automation promises to speed up production time, but it’s early days. At the end of 2026, Silana will deliver its first robotic cell to German sports and leisurewear manufacturer Trigema.

After a ramp-up period of several weeks or months, the company says its robotic cell should be able to separate and align pre-cut fabric panels, stitch together hems and seams, insert labels and stack the garment while computer vision keeps a careful digital eye on quality. In this way, tank tops can be assembled without human intervention between the fabric stack and finished product, though T-shirt production isn’t yet fully autonomous because workers must manually insert sleeves.

The potential time savings could be massive: Silana says its technology can automate at least 90% of the direct labor minutes in a standard T-shirt and reduce manufacturing costs by 82% in high-wage countries such as Germany or the US. (These reduction estimates compare Silana’s robotic cells to conventional manual production in the same country, rather than the much lower landed cost of mass manufacturing in Asia.

Silana says the calculation includes the system’s price, depreciation, energy, downtime and technical labor, and assumes that one operator will eventually oversee up to 10 robotic cells.)

The North Star for machine automated production is made-to-demand production with no minimum order quantities, says Hofmannrichter. But while the time and cost savings could be significant, whether this delivers sustainability savings is still up for debate.

Peer-reviewed research published in the International Journal of Life Cycle Assessment examined 15 potential cotton T-shirt supply chains using different locations, transport modes, overproduction rates and levels of automation. Climate impact ranged from 1.51 to 7.68 kilograms of CO2 equivalent per garment.

The researchers found lower emissions in automated scenarios, but only when companies also reduced overproduction, shortened transport distances, used cleaner energy and improved process efficiency.

One of the biggest factors is how close production happens to the end consumer, since a significant portion of the emissions in garment manufacturing come from air freight. The environmental benefits of faster sewing robots are limited if a brand uses the efficiency gains to manufacture more products or fly more garments to market.

Then there’s the labor impact to consider. Silana frames automation as a response to shortages of skilled sewing operators and an aging workforce, particularly across Europe and the US. But its model would inevitably replace repetitive sewing roles with fewer jobs supervising and maintaining machinery.

In sourcing markets where apparel supports millions of workers, this could be particularly detrimental to a just transition. Technology that lowers costs without changing the distribution of benefits may make the existing system more efficient without making it more equitable, says Michelle Gabriel, resident fellow at the Yale Center for Business and the Environment.

The rebound problem

Demand-led production confronts one visible form of waste — unwanted stock created by inaccurate forecasts — but it does not automatically reduce fashion’s total footprint. If a brand sells more of every style but introduces many more styles, efficiency per garment can coexist with rising production, resource use and post-consumer waste.

“While producing less product per design can increase sell-through, it can also lead to more designs,” says Karla Mora, founder and managing partner of Alante Capital, which invests in technologies intended to reduce fashion’s environmental impact, including MannyAI. The firm recommends brands begin with approved, preferably sustainable fabrics and design around them; set targets for how many times inventory should sell through and be replenished over a given period; and use AI tools to flag when assortment growth outruns proven demand.

Founder intent is critical, Mora says, because the same on-demand engine can reduce an established brand’s inventory exposure or power an endless sequence of creator drops.

Lu sees the same fault line in Shein’s model. “Reducing unsold inventory does not necessarily reduce total environmental impact; it may simply shift the problem from excess inventory to increased production, consumption and post-consumer textile waste,” he says. Better sell-through might be commercially valuable, but it is far from an adequate sustainability denominator on its own.

The Shein-Everlane combination crystalizes the stakes. Mora believes ultra-quick manufacturing doesn’t necessarily have to turn every user into a fast fashion company if the brand retains its standards for quality, materials, and assortment. Gabriel, however, is more skeptical, arguing that infrastructure carries the incentives of the model that built it, making a brand with less influence likely to struggle to resist a parent company oriented around speed, efficiency, and growth.

The outcome will depend less on whether Everlane can access Shein’s capabilities than on what it is rewarded for doing with them.

What speed is for

The cost-per-minute metric is unlikely to fade from current production calculations. Manufacturers need to know the dollars and cents required to take a garment from inputs to finished goods, and brands will demand competitive prices.

But a metric designed for one production stage is not the best way to judge the system as a whole. Forecast accuracy, full-price sell-through, inventory turns, working capital, supplier reliability, living wages, absolute production, recovered value and final disposition expose costs that a cheaper sewing minute can obscure.

The sustainability case depends on what companies do with the time they save. Faster systems can preserve value, but those gains disappear when greater efficiency simply fuels additional volume.

Faster approvals can prevent months of inventory exposure. Quicker replenishment can replace a speculative bulk order with a measured commitment.

An extra minute of refurbishment can rescue a garment before it loses its season. Conversely, shaving labor from a T-shirt serves little sustainability purpose if the gain funds more styles, tighter supplier demands, and rising output.

For Gabriel, evidence of a different operating logic would be visible in absolute rather than intensity-based measures: production volumes falling while revenue and margins hold or rise; supplier relationships becoming longer and more stable; wages improving on factory floors; and companies reporting the value created per customer over time rather than units sold.

Fashion’s obsession with speed, newness, and cheapness is often presented as an immutable feature of the market, but Gabriel rejects that premise: “They are choices, and naming them as choices is the first step toward making different ones.”