The Costliest Missing Signal in the American Food System

In 1996 I was sitting in a room full of people losing their minds about ecommerce. I was working in Information Systems and the certainty was absolute. Brick-and-mortar was finished. Within a year, we would all be buying everything from our couches.

We wrote a white paper that said something different: that the first thing people would reliably pay for online was subscriptions. Specifically, and I say this with complete professional seriousness, porn.

We were right. That argument ended up cited in a Clinton White House framework on global electronic commerce. I will let you sit with that for a second.

The point was never porn. It was the middleware. The payment systems, the fulfillment networks, the security protocols nobody was watching while everyone argued about the storefront.

That is what this piece is about.

A CPG R&D team spends eighteen months and serious capital reformulating for a GLP-1 consumer whose biological response to their product they have never measured. A grocery category buyer commits next season’s orders against last quarter’s numbers while the basket shifts underneath them in real time. A foodservice operator builds a menu cycle around a consumer whose nutritional needs changed the moment they started the drug.

I have watched versions of all three of these decisions get made. The people making them are not careless. They are working with the only signals the system gives them. The system has never been built to give them the right ones.

Three capital decisions. One consumer. No shared signal between any of them and the body driving all three.

And it is not just Apple. The data is sitting across an entire architecture of closed systems. Apple and Google hold wearable data. Insurers hold claims data that maps diet-related disease to purchasing behavior at population scale. Pharmacy Benefit Managers process every prescription claim in the country. Health systems hold clinical data from every GLP-1 trial ever run. None of them are in the food business. None of them have a systematic commercial relationship with the people making your food.

The wall is not one company. It is an entire infrastructure with no obligation to connect.

Before dismissing this as an analytics problem that better data vendors can solve, consider what companies like Tastewise and Spoonshot are actually doing: social listening, restaurant menu scraping, retail signals. Useful. Not the signal. Social conversation about protein is not the same as population-level glucose response data. The gap between what people say they want and what their bodies actually need is precisely the gap this piece is about.

The food industry has a $26 billion blind spot. Not a forecasting problem. Not a technology gap. A structural disconnection between the most valuable nutritional dataset ever assembled and the manufacturers, grocers, and operators who need it to make any decision worth making. That $26 billion is not what the industry is spending. It is the projected size of the market opportunity GLP-1 driven food innovation will create by 2034, according to market research firm Cognitive Market Research. It is what is at stake for whoever solves the connection problem first.

It is a structural problem. The missing link is not a better dashboard. It is infrastructure that has never been built. GLP-1 is the loudest signal right now. It is not the only one.

Europe is not solving this for food. It is doing something more consequential. It is building the architecture that makes solving it possible.

The European Health Data Space entered into force in March 2025, establishing a single EU-wide framework for health data portability across institutions, borders, and sectors. Patient records, prescriptions, diagnostic data, all of it moving through standardized interoperability infrastructure with the individual in control of access. Key operational provisions apply from 2029. Three frameworks — the Health Data Space, FiDA for financial data, the AI Act for enforcement — one regulatory philosophy: data portability by design, by law, with penalties for non-compliance.

The food application does not exist yet. But the infrastructure that would make it possible is being built by legislative mandate. In the United States it is not.

South Korea has launched Food QR, a government-operated real-time information service delivering live safety, allergen, and nutritional data at the shelf, with mandatory compliance being phased in through 2026. Japan is revising national nutrient labeling standards based on its 2025 Dietary Intake Standards. Both markets are pushing health information toward the consumer in real time. The American consumer receives a nutrition facts panel designed in 1994.

Every fragmented data problem in modern business has been solved the same way. Not by forcing the holders to share, but by making sharing commercially rational for everyone in the room.

Michael Bloomberg did not create financial data in 1990. It already existed, locked inside banks and trading desks that had every reason to hold it and no reason to open it. He built a terminal that made sharing commercially irresistible. Aggregated the signal, standardized it, sold subscription access to the people who needed to make decisions with it. Nobody owned the underlying data. Bloomberg owned the pipe.

The terminal became so embedded in financial workflows that subscribers could not leave even when they wanted to. It costs $27,000 a year per seat. Eighty percent of subscribers say they cannot do their job without it. The switching cost is not the price. It is the workflow.

That business did not exist before Bloomberg built it. The category had to be created.

The food parallel is exact. Fragmented health and nutritional data sitting in closed systems across Apple, Google, insurers, health systems, and pharmacy networks. Manufacturers, grocers, and foodservice operators making billion-dollar decisions without systematic access to any of it. A proven subscription business model waiting to be applied to a different data type in a different industry.

The neutral aggregator does not need to own the data. It needs to aggregate it, standardize it, and sell access to everyone who needs to make decisions with it. The Bloomberg terminal for human nutrition. The category does not exist.

So why has nobody built it.

The technology exists. The business model is proven. The need is visible to anyone paying attention. The reasons it remains uncreated are structural and they compound each other.

Apple and Google hold the richest longitudinal health data ever assembled. It is not a byproduct of their business. It is the business. Opening it, even through a neutral aggregator, weakens the lock-in that makes the watch worth buying. They have no incentive to share.

Insurers have monetized their claims data internally and in some cases sold it to pharma. Not to food manufacturers. Not to grocers. The commercial relationship has never been built.

The FDA drew its jurisdiction lines before this data existed at scale. Congress has introduced multiple bills that would require consumer consent before wearable health data can be sold, including the Smartwatch Data Act and the Health Information Privacy Reform Act. In June 2026 a House subcommittee held the first hearing on comprehensive federal data privacy legislation that would cover wearable health data. None of it has passed.

Underneath all of it is a coordination problem that makes Bloomberg’s original challenge look manageable. Bloomberg’s customers all wanted the same signal for the same reason. A CPG R&D team, a grocery category buyer, and a foodservice operator are asking completely different questions of the same data. Nobody has organized the demand side.

This is not waiting for better analytics. It is waiting for someone to decide the friction is worth it.

In America the regulatory path is moving slowly and going nowhere fast. Waiting for Washington to mandate the pipe is not a strategy.

The commercial path is already open.

Continuous glucose monitor companies like Dexcom and Abbott already operate commercial data licensing and partnership architectures independent of Apple — the same model that could be extended to food has already been proven in adjacent medtech and cardiometabolic research sectors. Pharmacy Benefit Managers already license anonymized prescription data to pharma. Grocery loyalty programs at Kroger, Walmart, and Target hold hundreds of millions of consented purchase transactions. Clinical trial data from GLP-1 studies alone is among the most detailed nutritional response datasets ever assembled for a single drug class, and most of it sits in academic databases with no commercial application yet.

The neutral aggregator does not need a single piece of legislation to get started. It needs a founder who sees what Bloomberg saw. Fragmented data. Willing sources. An industry making decisions in the dark. And a subscription model that makes sharing commercially rational for everyone in the room.

Every company spending eight figures on reformulation right now is making a bet. The company with access to biological response data at scale is making a decision. That gap is not closing on its own.

Seventy-one percent of CPG executives are already deploying AI, up from 42% in 2024. Every one of them is training it on the same inputs: purchase data, syndicated research, social listening. The AI is not the problem. The signal it is training on is. Build the infrastructure that carries biological response data into the food system and the AI gets smarter overnight. More than that — you have just created a category that does not exist yet. None of that infrastructure exists. Every piece of it has to be built by people who do not have those jobs yet.

The question is not whether this gets built. It is whether the food industry helps shape what it looks like before a tech platform, a pharmaceutical company, or a health system decides to build it instead.