We have put extraordinary intelligence into improving Instagram and the rest of the social-media economy. I think too much of our ambition has settled there. We should be asking what the same appetite for difficult problems could accomplish in manufacturing, equipment, maintenance, and the systems that keep the country working.
The country still needs people who can improve a machine, organize a production process, qualify a component, or make an old industrial system more useful. Those problems deserve ambitious careers and serious investment.
The reward is visible
Meta’s 2025 annual report gives a sense of the commercial scale. The company reported $196.2 billion in advertising revenue and $201.0 billion in total revenue, alongside $57.4 billion in research and development expense. Those are worldwide figures for Meta as a whole. They cover several products and kinds of technical work; they are not an Instagram development budget. [1]
The company describes artificial intelligence supporting content ranking, recommendations, advertiser tools, generative products, and internal productivity. It also invests in longer-term computing projects. An advertising business operating at that scale can support enormous technical effort. The commercial reason for improving discovery, engagement, and advertising performance is easy to understand. [1]
My question concerns the outcomes we choose to reward. A feature can be commercially successful because it helps sell more advertising. That success does not answer how much additional attention the feature deserves from the people and institutions shaping the country’s future. Private returns and the productive capabilities a country needs are different measures. We should be willing to discuss the distance between them.
I consider part of this effort a waste of scarce talent and attention. A profitable feature can still be a poor use of our collective ambition when its main achievement is keeping people scrolling. That is a judgment about priorities, not a calculation of factories lost: Meta’s accounts cannot tell us which engineers would otherwise have chosen industrial work. We should give people a more convincing opportunity to pursue the physical problems around them.
The work that remains
Consider the questions inside an ordinary industrial operation. How can a process use less material? What causes a recurring failure? Can a replacement part be identified correctly before someone orders it? Can a product be made consistently enough that every customer does not require an exception? How can a small manufacturer understand its capacity and commitments?
These are development problems. They require observation, experiments, design, information, and the ability to learn from a result. Some call for new machinery. Others call for better software, clearer records, a different component, or a more reliable handoff between departments. The outcome can matter for years to the people doing the work.
Older technologies still have problems worth solving. Hydraulic systems, mechanical equipment, and existing production machinery need people who understand their applications and can improve them. A newer technology should earn its place through what it does for the job. An existing one should receive continued development when it remains useful. The installed equipment base deserves engineering attention throughout its working life.
Software can build productive capacity
I use technology to connect information and understand industrial questions. I want more of that work. Useful industrial software can improve the connection between a design and a production process, help diagnose a fault, or make the history of a machine accessible to the person maintaining it. The purpose of the development should remain clear: what can the user now do better?
Deere reported average herbicide savings of around 50% from See & Spray in 2025. Its reported results illustrate digital development serving physical production; actual savings depend on operating conditions. [2]
A programmer working on a machine and a programmer working on a social platform may both face demanding technical problems. The question is what the work makes possible. I want the industrial problems to attract the same curiosity, professional respect, and willingness to finance a difficult development period.
Recognize what industry already invests
Industrial research already receives substantial investment. NSF and the Census Bureau estimate that manufacturers performed about $394 billion of research and development in the United States in 2023—55% of domestic business R&D. That is substantial industrial effort, spread across very different activities. [3]
It strengthens the case for asking better questions about what we develop next. A national total cannot tell a particular equipment builder whether the process, supplier, or technical skill it needs is receiving enough investment. Nor can it tell a young person whether there is a visible, well-supported career working on those problems.
Meta’s worldwide 2025 accounts and a domestic industrial survey for 2023 are not competing entries in one budget. They help describe two parts of the economy. Deciding how to support more useful industrial work requires looking at the specific opportunities, the people who could do it, and the conditions that would let them succeed.
Give useful work a credible path
I would like to see industrial employers make development work more tangible: a defined problem, access to the equipment and people who understand it, paid time to test an improvement, and a way to share in the value created. Training should connect to work someone will actually perform. Commercial teams should bring recurring customer problems into product and process decisions.
Investors and business owners can ask how a project expands capability over time. Does it make production more dependable? Does it reduce an input? Can more people operate or maintain the system? Does knowledge stay with the organization when an experienced person leaves?
Instagram will continue to attract investment for understandable commercial reasons. I want America to show greater ambition for the rest of the work. We have people capable of solving demanding problems. More of them should have a convincing opportunity to spend their careers making the physical world work better.
Make the industrial problem visible enough to fund
A recurring equipment problem needs a development brief as concrete as a software feature: who is affected, what prevents useful work, which evidence identifies the cause and what would count as an improvement. For a hypothetical replacement-parts project, that might mean resolving conflicting identification records before building a more elaborate ordering interface.
The industrial opportunity can begin with a modest, bounded result. Reduce a documented source of rework, preserve an approved configuration or make a repair history usable to the next technician. Then examine whether the result can be repeated and supported. That is the kind of development path I want more people to see: a difficult real problem, access to the people who understand it and a clear connection between effort and productive capability.
Sources
- Meta Platforms — 2025 Annual Report. Year ended December 31, 2025. Revenue and R&D figures are worldwide totals for Meta, including multiple products and kinds of technical work; they are not Instagram-only spending.
- Deere & Company — Fourth Quarter 2025 Earnings Call. November 26, 2025, page 14. Management-reported See & Spray acreage and average herbicide savings provide an example of digital technology supporting physical production.
- NSF NCSES — Business R&D Performance in the United States Increases to $722 Billion in 2023. Published September 29, 2025. Table 2 reports manufacturing's share of domestic business R&D for 2023; it is not directly comparable to Meta's worldwide 2025 accounts.