The Rise (and Fall?) of Food Robotics, Part II
The Rise (and Fall?) of Food Robotics, Part II
The Challenges Facing Restaurants and Robotics
Welcome back to our series on the current state of Food Robotics. While in Part One we took a brief look at the history of restaurant innovation as well as the current “winners” and “losers,” today we’ll continue that exploration by looking at the challenges facing both the robotics industry and food and beverage sector to understand why we haven’t seen the robotic takeover some predicted and ultimately answer the question: If robots are supposed to become commonplace, why haven’t they yet?
What’s Cooking in Restaurants
To be clear: It’s not like restaurants are sitting on their hands in the face of high turnover, upped labor costs, increased cost of goods and the loss of lower income customers. Many of these companies are publicly traded and have been hammered on their quarterly calls over what they are doing to solve these issues. And in large part, the companies are listening and are investing in automation among more traditional approaches. Other than a few concepts from McDonald’s and Chipotle, the automation has largely been digital. As I mentioned in the previous article, ordering kiosks are now commonplace in almost every fast food restaurant and store layouts encourage customers to use these kiosks. Some only have a traditional POS system “just in case” and kiosk usage is effectively 100%. This frees up the cashier to now focus on output: prepping, making and delivering orders. It doesn’t necessarily reduce labor, but reallocates to the next critical operational chokepoint.
This gets accelerated with the introduction of artificial intelligence. Many restaurant chains have adopted some form of AI order-taking at their drive thrus. 2023 was a banner year for SoundHound and Google’s AI and there’s no sign of it slowing. Frankly, this is great! If you’ve ever spent any time at the back of a restaurant with a drive thru, all the action is with the person running that window. They are taking and confirming orders, calling out needs to the rest of the staff, verifying items in the bags, filling drinks, collecting payment and handing off the meals. It is an incredibly stressful and demanding role; no wonder turnover is so high.
If we extrapolate what automation and AI can do (without robotics,) we see some amazing things! Adding a few computer vision cameras and some complex models, AI can move beyond order taking and will eventually do order accuracy, food quality assurance, customer satisfaction and inventory. And there are companies out there already trying to solve these problems. These will definitely provide value to restaurant operators, but at the end of the day it does little to increase throughput. Without physical automation and robotics in the kitchen, throughput will always be a simple math equation involving square footage of the kitchen and number of employees that will fit in it. While we’re seeing restaurants lean into technology with these AI solutions, we’re not seeing many investments into hardware. This might be because hardware and robotics aren’t without their own challenges.
Hardware Is HARD
Having spent my entire career in the development of hardware, there has always been the recurring phrase around the proverbial watering hole of hardware companies: Hardware is HARD. Why is that? In short: compared to opening a restaurant or creating a new piece of software, robotic development is slow, expensive and risky (more on this comparison in a minute). And implementing robotics in a fast-paced restaurant setting, there are a few critical challenges for that application. Hardware development is expensive, but more specifically there is an affordability gap at the unit level. While continuing to decrease in cost, robots are still generally more expensive than what most restaurant owners are ready to invest in capital equipment. Along those lines, implementing a robot doesn’t generally reduce labor costs unless it removes an entire person for a whole shift. Otherwise, operators are moving that person to another part of the store. In this case, the ROI has to come from increased revenue, which is difficult to directly attribute given every store is trying a thousand things simultaneously to increase revenue. The other arguable benefits of robotics: consistency, reduced waste and speed of service can be hard to attach a specific monetary value and thus hard to make a clear ROI case.
The other big thing is space. Restaurant kitchens are designed to squeeze use out of every square foot for humans to safely use and move around, not for automation. As operational space decreases, technical complexity goes up, which impacts reliability and cost. To add to that, variability in restaurant layout and equipment also requires the design to be flexible enough to customize to each install. And then humans still have to work around this thing! So robotics may be able to help labor and throughput challenges at the restaurants, but aren’t a sure bet. So what are the biggest chains betting on?
Always Follow the Money
Every year, restaurants invest significant funds to increase their revenue and market share. And boy, are they spending. From what I can find in the public sphere, McDonald’s is leading the pack with spending over $2.4 billion in capital expenditures. The bulk of this is going to “tech” which seems to mean moving all their software to the cloud and implementing AI. Burger King is also in the ten-figure club, spending over $2 billion to update restaurants. Wendy’s is taking notes and spending $100 million on advertising and digital upgrades. Chipotle seems to be bucking the trend by investing $100 million into the future of food. This includes some robotics companies like Hyphen and Vebu, as well as some companies working on meat alternatives and a more sustainable supply chain.
And, look, investing in growth, infrastructure and advertising are of course worthwhile. Doing things like updating stores, optimizing processes and improving loyalty are tried and true methods in business. All the marketing, loyalty programs and updates might increase demand, but if you can’t keep a team onsite that can respond to that demand, you end up with slower speed of service, decreased quality and employee burnout. It can also result in missed revenue. I’ve heard stories of one large chain turning off digital orders during lunch because even with a “full” staff they couldn’t support the order volume. They are running out of things to optimize outside the physical kitchen. I would argue they are playing an old game, but the rules are changing. This is where we need to talk about robotics and the cost of development.
What Does It Cost?
With any new venture, there are a few key milestones that indicate success: product-market fit, revenue, profitability and ROI. So, let’s take a look at how and when these milestones occur in restaurants, software and hardware.
While starting a restaurant is by no means easy, it is pretty straight forward: find a location that hopefully likes your cuisine and has the foot traffic to support a good business (product-market fit), build it out, create your menu and supply chain, hire a staff and boom! You’ve got a restaurant. Once the restaurant is open, you can now get revenue in the door. And a single unit restaurant can become profitable within a year and, if they really nail it, the ROI is three to five years on that initial investment (on average $250k - 500k per restaurant). Heck, you might even be able to just buy a space already equipped for a restaurant, whip up your menu, and get to work quickly which can make that ROI even better! Those initial capital expenses and risk can be high, but if you did your math right and nailed your market, you can rinse and repeat to expand.
Software products may not cost as little as a restaurant, but with no capital equipment to purchase, initial investment risk can be significantly lower. Like a restaurant, it still needs to hit product-market fit, but requires scale to be profitable. A simple piece of software can be built and tested in a matter of a couple weeks or months. Iterations cycle at the speed of coding to get to a shippable product and then validation with product-market fit testing can begin. Scaling from 1,000 to 10,000 to 100,000 users can happen quickly, drawing revenue while the product is in its infancy and startup costs are low. As it grows, changes can still be made, fixes and updates to that product can be deployed without needing to slow down or start over. Start up costs can vary widely, but can be non-existent to a couple million dollars. Scaling can cost more, but the initial scale is in the seven figure range. Arguably at this point the company will have customers, revenue and notional product-market fit with path to profitability (if not profitable already). And the ROIs can take years, but can be astronomical. As an extreme example: Squarespace was built in a dorm room by one kid and within two years he was making $1M yearly in revenue. It’s now worth close to $5 billion.
To build a medium complexity piece of hardware (let’s say, a smart combi-oven that can load and unload its meals), you need to design it, build it and test it. Product-market fit is largely hypothetical based on external data that informs the design. Design and building requires a variety of mechanical, electrical and software engineers who work together to get a prototype built. This whole process can be three to four months. It is then repeated as they iterate on the design, working to build a fully working device. This can take a year or more and can easily cost one to five million dollars. To scale that product can take another year at least, costing millions of dollars more. And this is all before you’ve made a single dollar on your product! In short, building a hardware product can take a number of years and tens of millions of dollars. It takes years and millions of dollars to get revenue, much longer to make a profit and longer still to get to ROI.
This is the crux of the challenge for restaurants and automation. There are orders of magnitude difference in when revenue and profitability happens. The average startup cost to get to revenue for restaurants is six figures, software is seven and hardware is eight! No wonder we’re not seeing the restaurant industry invest in hardware. Even with a company like Wendy’s investing $100M, they are going to see almost immediate returns. $10M is a lot of money for restaurants and having to wait several years for a non-guaranteed return? That’s a hard burger to swallow.
A related challenge is who has the money and who benefits. For a lot of the big chains, they are heavily franchised. For some revolutionary piece of restaurant robotics, a franchisee could see direct benefit to their restaurants with its implementation, but they are also the most cost-sensitive. Replacing a commercial fryer for a thousand dollars is a big expense for them. Signing up to lease a robot for a few grand a month? They may not be able to afford it. The parent companies have a lot more cash on hand, but won’t see the direct financial benefit of investing in robotics, so they aren’t as willing to put down the money.
So What’s the Point?
We’ve seen how restaurants are finding ways to move all labor to the kitchen and automate or remove (RIP dining rooms) the rest. While this transformation hasn’t completely finished, it will soon. This leaves the kitchen as the sole bottleneck. Throughput is 100% reliant on human labor, which is becoming more costly to the company and more stressful to the employee. They have maximized humans’ ability to operate in a tiny area and put all their eggs in that basket. If labor challenges continue to increase (which is expected), you’re gonna get some broken eggs. Restaurants will need to start automating the kitchen, but aren’t willing to put big money there yet because they are still hellbent on optimizing everything else first
Development of robotics is a slow, expensive process. The robotic companies need to have funding to get through product market fit and then scale. The traditional funding channels, namely venture capitalists, aren’t investing in the space like they were just a few short years ago. So how do we solve this problem? I'll address this in the next article and work through where we go from here.
Contributed by Darian Ahler.
Darian Ahler has spent two decades in the food automation and restaurant robotics industry. He most recently served as Head of Product Strategy at Vebu (formerly Wavemaker Labs) and CEO of Bobacino. Prior experience includes co-founding Truebird and work at AeroFarms.
