Perpetual AMA for Caleb


@Caleb I am trying to understand a bit more in detail the economics and efficiencies of running large scale urban farms using hydroponics in terms of CO2 emissions and electricity costs compared to traditional farming, in particular in northern countries such as Scandinavia where there is a lack of sunlight or in countries near the equator with lack of water sources.

I have found some scientists in universities such as Cornell or in Utah where they seem to criticize Vertical Farming claiming that the numbers do not make sense (, for example).

Do you have any particular reliable research or numbers you have gone through to be able to show truly the benefits of urban farming. I have run some numbers from hydroponic systems suppliers in the US and Europe and with their electricity consumption and the CO2 emissions of the power plants in the Scandinavian countries and it is comparably much lower co2 intensive than transporting those same good from countries in Southern Europe or even Africa or South America.

Would be great to have some input on that. Thanks. Albert


@al_payi you are asking me the as yet unanswered question - that is why we are here doing the work to build out the economics by setting open standards for hardware and data sharing. It is near impossible for anyone at current time to answer that question definitively, even dare I say it, cornell :stuck_out_tongue:

That being said here are some things to think about:

  • In the Short Term (IMHO and I love to be proved wrong)
    Vert farming will not be economically viable for lettuce, culinary greens, eat fresh pick fresh vegetables (just google farmed here, podponics, verti-crop for starters bannnkkkrupppt)
    Vert farming will be economically viable for pharmaceutical/nutraceutical natural derivatives, bio-active compounds (think cosmetics) and natural flavor derivatives (low volume, high margin)
    Greenhouses are part of the control environment agriculture umbrella (just like vert farms) and are definitely going to play an increased roll in crops mentioned above
    Full control environments (growth chambers like conviron and PFC) will be useful in the interim for research for greenhouse and vert farming experiments

  • Caleb you just broke my heart why dont vert farms work now?
    Capital equipment cost per growing square foot is incredible high and people are scaling unproven systems MUCH MUCH too fast. When you scale fast your failures are trememdous
    The actual growing square foot multiple (defined as the total number of growing square feet **minus **aisle ways, mechanical space, packaging, harvesting, etc / total square feet of building) is too low, believe it or not some of the facilities Ive seen with 6/8/10 layers which ostensibly seem to have a 6/8/10X multiple, once you do the math have a 1.2-1.6 multiple (1.6 being the best Ive ever seen in US/Japan/Europe)
    Labor costs are too high ever wonder how most vert farms get harvested now-a-days? a laborer…on a cherry picker…insanity
    Heating/Venting/Air Conditioning (HVAC) is a silent killer of vert farms ATM. People thinking its a wonderful idea to use an old warehouse, then realizing what they really needed was a heated and cooled data center - not a leaky warehouse. Its the number one operational energy expenditure for vert farming

Here is some moooore bad news for folks like us that want answers:

  • “whats the energy cost in vert farming vs convention”…a seemingly simple question on its face but once you start playing the game of following the shrinking kilowatt you will quickly run into dead ends.

Specifically -

  1. the energy equation of a piece of fruit in a grocery store is dynamic it changes wildly depending on what time of year it is, where in the world you are and what the piece of produce you picked up happens to be. In fact most grocery stores sell produce at a loss, you heard me right, they pretty much pay to give it away to get you into the center of the store to buy the higher margin items.

  2. There is no easily accessible data on energy consumption along the dynamic chain that brought your produce to you.
    *The seed, presticide, mineral etc companies arent reporting the energy used to create the commodity inputs
    *The distributor of those commodies isnt reporting what energy they used to get it to the COOP/Farm
    *The farmer isnt collecting and reporting energy used for tractors, pumps, irrigators, etc
    *The cold chain distributed isnt reporting data on their refer-containers used to take the produce from the farm to the pack house
    *The pack house isnt reporting data on energy use
    *The distributor from the pack house isnt reporting data on energy use
    *The Retailer isnt reporting data on energy use
    *And I’ll bet you arent reporting data on the energy you used in the last mile of getting your food from the grocery store home (Car, refrigerator,etc) which can sometimes be the worst of them all

So then you might ask the question…with all this energy use in conventional farming…how does it all work economically?

  • Enter SUBSIDIESSSSS Farming of any kind would not work without HUGGGGE subsidies (*Farm Bill in the US) and currently there are no subsidies being applied for vert farming…

All that being what it is…we still must try. There are smarter people than me working on aspects of this (*esp carbon footprinting) but I had a simple idea. Well if we have no access to the data of conventional farming and it is dynamic anyway… maybe we could create some metrics for vert farming (as I call it food computing) that are irrefutable and unbiased and this is what I came up with

  • KWH per Gram (of Product)

  • Liters (of water) per Gram

  • Labor per Gram

  • Amortized capital equipment per Gram

Then we can link those metrics to the gram value of the particular product. For example. about 4 years ago we ran some simple version of this where we calculated all the energy used in the food server 1.0 farm prototype at MIT (KWH) and multiplied it by the rate we pay for energy at MIT which is about 0.14 USD per KWH. Then we measured the total weight (Grams) of the harvest from one cycle of the same crop (buttercrunch lettuce). We divided the two to get the Total Price of Energy of all the grams of lettuce and then divided by the number of lettuces to get the price of energy for that particular lettuce. (Im writing this on a plane, I think that all makes sense, but you might need to interpolate and my fingers are getting tired). OK so what we found was it was abou 0.60 USD for the average 300 gram head of lettuce for all energy in (which in that case included HVAC, lights, pumps, etc). The cost at the supermarket (you have to be specific in supermarkets - there are value markets and quality markets - Im speaking of value driven markets *ie not whole foods ;)) for the same 300 gram head of lettuce was 2.19 USD. So on initial assestment (*of course omitting labor, equipment, water) we thought our energy use was actually not so bad. But then again, the economics of grocery produce are as I stated above, complicated.

I started watching the video you mentioned. I also started writing a litany of notes on his comments. Then I erased them because…well…I do not find a lot of his talk interesting/progressive/intriguing and I also was not finding my comments particularly in line with how I like to engage with science/invention/thinking/making and very similar to a lot of trolls in the youtube comments. So to his video I simply say…

“The horse is here to stay…the automobile is a fad”

and instead of dwelling on that video…Id encourage you to watch an old ted talk I recently dug up from Danny Hillis. Danny was one of the inventors of the internet, he made something impossible, possible. I respect people that can do that with rigor, passion and by suspending disbelief. A salient point that speaks directly to how I feel about vert farming, “We are the amoebas and we can’t quite figure out what the hell we are creating. I think that there really is something coming along after us. I think it is very haughty of us to think we are the end product of evolution. I think all of us here are part of producing whatever that next thing is”

Commercial scale of food computer
Building a food computer on developing communities
Growing Veggies packed w/ Flavor & Nutrition

@Caleb Awesome post! Have you published any of the food server energy and price calculations you mentioned? That’s super interesting.


Nope they were proof of concept in a leaky farm with bad software :wink: Thats why Im building the new lab in middleton! thats what all this work is about @wsnook !


I didn’t realize that’s what you had in mind with the new lab. Very interesting. @rbaynes fyi, I’m interested in this angle of things. I’m curious about ways I might be able to help and about how to tie the food server work back to the MVP stuff so the software is reusable.


Thanks @Caleb for the detailed and rigorous post on my comments and sharing this great TED talk. It has given me some great perspective on your position towards the economics and sustainability of vertical farming and I also learned many things that I for sure didn’t know about.

With regards to the comparison between vertical and traditional farming I have also started to develop these metrics on already existing systems (BrightAgrotech, UrbanCrops…) using the data available on their website and through given brochures… and the results were quite similar one to the other:

UrbanCrops: 7-15 kwh/kg produce (depeding on if its lettuce or basil) which is around 0.3-0.6 USD/lettuce.
BrightAgrotech: 8 kwh/kg

So it seems results are consistent to those obtained in your CityFarm. Water usage was also quite similar for both and around 6-10 L/kg of crop and if using Sweden as the base for calculations (we are quite carbon free :P) the CO2 was around 0.5-1 kg CO2/kg product. So taken into consideration all the labour, equipment costs… there is still at least 50% margin on the final price in Sweden.

Definitely Cornell and that video i attached… not the best sources for encouragement… these are just my source of motivation to prove them wrong.

I would also like to bring to you another topic. I am starting my own hydroponics venture to sell greens and vegetables directly to the consumer market in Stockholm (hopefully i can also prove you wrong and not end up in bankrupcty…:S). In that endeavor I would be very interested in sharing and collaborating with OpenAg but i am not sure if these collaborations are possible and how would they have to be developed. In that effort, my goal is to start with a small prototype (50-350m2) and at the same time develop our own OpenAg research in KTH (my university) building first a PFC v2.0 and later on working on developing our own food data center. We have gathered already potential KTH Researchers interested in the working on CEA and it would be great to hear your feedback on that.

Finally as I have seen that you will be presenting in Stockholm for the food forum in June, is there a way to arrange a short meeting to discuss this a bit further?

And final final question, is there a way for arranging short-term stays at your MIT Lab (for students at other technical universities for example during the summer?). This could be a great initiative to increase the OpenAg community throughout the world.

Thank you for reading and love to hear your thoughts.

have a great weekend.



Hi @Caleb

I am an agricultural scientist and microbiologist who currently works in public health and food safety in Australia. I have recently developed an interest in sustainable food production and urban agriculture as a novel way of satisfying the requirements of feeding a growing global population. As part of this, I have been reading extensively on the OpenAg project and hope to commence construction of my own food computer in the near future to better contribute to the community.

My question is more one that is focussed towards the work that I currently do, particularly around food safety and public health. As you may be aware, traditional horticultural produce is responsible for a significant amount of global foodborne illness (e.g. the E. coli O107 outbreak in Europe during 2011, the Jensen Farms cantaloupe outbreak also during 2011). A large proportion of this produce is consumed with little or even no preparation and pathogens such as Salmonella, Listeria, norovirus and E. coli associated quite readily with this produce can cause severe illness, particularly in vulnerable individuals. Additionally, there is quite an amount of research going on in the field of plant/human pathogen interactions, particularly in the area of internalisation of pathogens in horticultural produce, and the jury is still widely out as to how to guard against these potential routes of exposure.

Whilst there is no doubt that there are significant nutritional benefits for public health associated with fresh home grown produce and the risks associated with long supply chains are largely negated, food safety is often overlooked in home grown produce. I searched through the forums but I haven’t been able to locate any discussions surrounding food safety and the PFC. My question is: Has the OpenAg project had any discussions or given any consideration to implications for food safety when using production systems such as the PFC?

Thanks for your time - I would be more than interested in hearing your thoughts.



Hey @andrewW the short answer is no! Its pretty crazy all these CEA projects are spinning up without much thought to food safety! (containers, factories, etc). In fact there is no USDA certification process I am aware of so its up to the developers. You sound like you have really special and useful knowledge on the subject. Why not start a thread in the forum around it and start imparting your wisdom? This forum is meant to be a place people come together to collectively problem solve and share knowledge - Feel free to call me out in your posts so I can learn too



Thanks for your reply @Caleb. Great idea and definitely something that I will get going in the community forum.

Unfortunately, government regulators all too often reactive in these sorts of instances. Instead of exploring novel food industries, engaging with them to work collaboratively and ensuring food safety is built in to their development processes, they wait until something has gone seriously wrong and then regulate the hell out of it. We have fewer issues we can park the ambulance at the top of the hill so to speak and I am lucky to be involved with a food safety regulator that embraces this philosophy.

P.S. I will be visiting Boston for work in mid-July and it would be great to link up and have a chat with your group about this.


Hi @Caleb Harper.
I’m Jin Hyuck Lee in Korea

I’m Korean Polytechnic University student
I’m majoring in mechanical design
I am crazy about Food Computer

so I made a start-up club recently and gather my crew

there are 10 people who want to build PFC

They are competent students and workers in each field.

i have a plan to supply PFC in korea

and build a web site to service about all of PFC ( recipe, expendable things , korean community etc)

and finally we can share korean unique recipe (i really hope so)

we are really care about safe food in urban and sustainable agriculture system

we are starting to hardware parts in Digital blacksmith in Seoul now

few days ago, i went to urban agriculture forum hosted by a public agency

and i can have a chance contact with the mayor of Siheung.

Go to the main thread

can i ask you for help in any way.

to convince business plans or just your message to officially

or objective data or even keywards to google it


if you want to receive business mail

please reply me and really thank you!


@Caleb do you have entry tickets for Nerdfarmers in EAT forum? I would love to come and hear you talk :slight_smile:


Hey!!! I have no idea…hmmmm…want to just show up at the Clarion Sign hotel tomorrow and we can try? Im a big fan of begging forgiveness instead of asking permission :wink:


Hey @Caleb that would be awesome. What time should we meet? Thanks, Albert


@Caleb Hello. I enjoyed your EAT Forum talk. It’s here for anybody that hasn’t see it.

With my food computer prototypes, my focus has been on reducing cost. I am curious what I should be aiming for to improve viability of food data centres. AeroFarms appears to have been built at a cost of >7000USD per square metre.

What is the cost of your food servers per m2 of growing area, and what is the target cost for them and the food data centres?


@Caleb Any chance that you could share more info/code/papers on your “Plant AI: Bayesian Optimization and Gaussian Process Regression” slide from the EAT Forum? It’s a bit hard to read the slide in the video. It looks like you are comparing three different lighting systems on one axis, while the other two are illegible. Based on my read it looks like you have 5-6 actual data points (green and gray points), while the others are just estimates from the model. How did you tune your hyper parameters with so few data points?

As for the methodology for effective modeling of ~3.2M data points per plant, how are you preventing over fitting? Are you using a time dependent PCA to shrink your search space? How do you account for different environmental conditions needed at each plant life stage in your models? Have you found a way to simplify how you think about plant recipes or found any differences in aggregating information? For instance, does applying a total of 600 W*hrs/day to a plant differ matter if the light is 50 W for 12 hrs vs 100 W for 6 hrs? If it makes no difference, then you could think in terms of total light allocated. However, my guess is that you can’t do this since there is an upper bound to the rate of reaction that a plant can perform.



@Caleb I wrote a post on another section, not sure if you are able to help here but it would be awesome. We just got the funding from KTH to build and develop a food computer :slight_smile:



Another question. What is the purpose behind the name OpenAg? I guess, how much Open Agriculture innovation are you hoping to foster here on the forum? I ask because the community and the discussions themselves are quite limited at this point in time to topics relating to PFC and specific growing conditions like hydroponic growing.

Are you looking to foster openness in all areas of agriculture? I personally would love to see the forum grow beyond where it is today and also include discussions and a community for regular plant breeding and/or regular growing even without the PFC or artificial environments. But maybe that’s just me. I really don’t know who is really interested in that kind of stuff on this forum since it has not been talked about much at this point.


@Caleb I am visiting a very good friend of mine from MIT at the end of February and I was wondering if it is possible for me to meet the you and the team even if its just for 20 min to see around your facility in Cambridge?
Thanks Most likely between 26-2 march?