Source: Youxin Newin
In the digital age, overwhelmed by online advertisements and consumerism, people only see the surface of Silicon Valley's tech boom. The tech elites loudly proclaim the grand slogan of 'creation' but rarely ask, 'what exactly should we create?'
In their 2025 work 'The Republic of Technology,' Alexander Karp and Nicholas Zamiska use an aggressive tone to expose the veneer of warmth in the tech circle: the soul of engineering culture is gradually disappearing. Only by abandoning the hollow 'soft beliefs' and reshaping the foundation of national defense and industry through hardcore technology can one survive in the turbulent geopolitical game.
This is by no means a lukewarm business inspiration book, but rather a hard-hitting manifesto about technology, power, and future warfare. Karp asserts that software is not merely a commercial tool but the ultimate weapon that determines the fate of nations. It reveals how core decision-makers across the ocean are attempting to reshape the 'state-technology' contract and return to realism with a more aggressive stance. It also forces every tech professional to ponder: when technology has the power to determine the course of civilization, what exactly are we innovating for?
From 'Dopamine Economy' to 'Survival Engineering'
Silicon Valley is undergoing a profound identity crisis. When we talk about the tech industry, what often comes to mind are social media, short videos, food delivery platforms—those consumer products meticulously designed to stimulate dopamine release.
However, understanding the early development of Silicon Valley reveals a completely different picture: Fairchild manufactured reconnaissance equipment for the CIA, Lockheed produced naval ballistic missiles in Santa Clara County, and DARPA funded the prototype of the internet. Innovation during that era served national missions and industrial-grade challenges, rather than trivial demands of the consumer market.
This shift was not accidental. Over the past two decades, capital, talent, and media attention have almost entirely flowed into the consumer internet sector. Venture capitalists chase the next social media sensation or e-commerce platform, while top engineers dedicate their careers to optimizing ad click-through rates. In this environment, a vast 'dopamine economy' was created: product design revolves around instant gratification, business models rely on user engagement time, and innovation has been reduced to capturing attention more precisely.
The problem, however, is that when the brightest minds are focused on how to make people spend five more minutes watching videos, who will address the critical issues that truly shape the future? When $Palantir (PLTR.US)$ engineers were debugging software on the battlefield in Afghanistan to counter IEDs, Silicon Valley’s dominant narrative revolved around Zynga's FarmVille and Groupon's group-buying frenzy. This is not merely a matter of industry choice but rather a systemic decline in innovative ambition.
Behind this decline lies a severe distortion in capital allocation. Data shows that in 2024, the U.S. Department of Defense's budget request for AI accounted for only 0.2% of the total defense budget, while financing in the consumer internet sector during the same period was tens of times higher.
The prosperity of the consumer internet masks a harsh reality: we are trading long-term strategic advantages for short-term commercial returns. Employees at Google once protested and ultimately led the company to withdraw from the Department of Defense's Maven project, citing the motto 'Do no evil.' However, in practice, 'Do no evil' has evolved into 'Do not participate'—avoiding any public projects that might provoke controversy and shirking responsibilities beyond commercial boundaries.
This absolutist view of 'technological neutrality' may lead to a lack of strategic responsibility. Silicon Valley's tech elites benefit from the dividends of education systems, infrastructure, and legal environments built through national investment, yet they may confine their innovations entirely within the framework of commercialization, failing to adequately reciprocate the system that made these achievements possible. A significant number of top technical talents lack a tangible sense of urgency regarding geopolitics and national security, preferring instead to avoid controversy and cater to consumer demands rather than engaging in more fundamental public domains.
Even more concerning is the misallocation of talent. David Graeber once sarcastically remarked: Where are the flying cars and Mars colonies we once dreamed of? The reality is that top talents are developing mobile applications, social media platforms, and food delivery services. This misalignment is not a market failure but an overextension of market logic—when everything is reduced to commercial returns, fields that cannot quickly generate profits but are crucial naturally become marginalized.
Companies like eToys, the 'Lost Toy Kingdom,' exemplify this state: we possess unprecedented technological capabilities but use them to create toys. This reflects the typical mindset of the internet bubble era: ambitious desires to 'disrupt' traditional industries, yet with the ultimate goal of selling goods more efficiently. This shallow ambition is becoming the default setting for the entire industry.
The advent of the AI era places humanity once again at the crossroads of engineering and ethics. Yet our discussions remain fixated on questions such as 'Does AI have consciousness?' while ignoring the more pressing realities: Who will control these technologies, and for what purposes will they be used?
Transitioning from the 'Dopamine Economy' to 'Survival Engineering' does not negate the value of the consumer internet; rather, it seeks to rebalance innovation priorities. We need to redirect a portion of capital, talent, and attention from consumer sectors toward areas vital to national survival, such as defense, energy, and healthcare. This is not a moral appeal but a strategic necessity.
Such a shift requires innovation in business models. Palantir's collaboration with the military demonstrates another possibility: technology companies can maintain commercial independence while serving public missions. The key lies in redefining the criteria for 'success.' It should encompass not only commercial returns but also contributions to national security, technological sovereignty, and strategic advantages.
VCs also need to rethink their investment logic. When the marginal returns of consumer internet diminish, should part of the capital be allocated to fields with longer cycles, higher risks, but greater strategic value? This is not charity, but a long-term business decision. Companies that can address 'existential-level' problems may ultimately create more value than consumer applications.
The shift from the 'dopamine economy' to 'survival engineering' essentially represents a paradigm shift in innovation: from satisfying individual desires to addressing collective challenges; from short-term commercial returns to long-term strategic investments; from technological escapism to assuming responsibility. This is not an easy choice, but it may be the most important one of this century.
The technologies that determine how we live and survive should not be reduced to consumer products. They require grander goals for guidance, firmer commitments for support, and a more sober sense of responsibility to navigate. The starting point for all this lies in how we understand the true meaning and historical mission of 'innovation'.
▍The Coupling of Government and Corporate Power in the AI Era
AI technology is reshaping the global power landscape, and the reconfiguration of government-enterprise relations has become a key variable determining national competitiveness. This is not just a Silicon Valley story but a strategic issue that all technological powers must face. When algorithms can determine the outcomes of wars and influence economic lifelines, the traditional boundaries between tech companies and governments begin to blur, sparking a profound global contest over power, responsibility, and governance models.
The strategic nature of AI dictates that it cannot be fully privatized or completely nationalized. Training large language models requires massive amounts of data, computing power, and capital, often with the backing of state resources. More critically, AI applications in military, intelligence, and critical infrastructure directly impact national security and geopolitical landscapes. When a technology can determine a nation's survival space, it ceases to be a purely commercial product and becomes an extension of national capability.
This transformation has triggered intense conflicts globally. As competitors rapidly advance in areas like AI militarization, drone swarms, and quantum computing, excessive adherence to commercial ethics by tech companies may become a national disadvantage. Data shows that some countries allocate far higher proportions of their budgets to AI defense applications than traditionally recognized, and this gap is affecting the global technological balance.
From a commercial perspective, such resistance also appears shortsighted. While the commercial application of AI technology is undoubtedly important, truly large-scale, high-value applications often originate from the public sector. Deployments in healthcare diagnostics, urban governance, national defense, and energy optimization not only represent vast market opportunities but also create long-term technological moats. Refusing to collaborate with the government equates to voluntarily abandoning these critical markets.
However, the challenge lies in the fact that government-enterprise collaboration is not a simple commercial transaction. It involves sensitive issues such as data sovereignty, algorithm transparency, citizen privacy, and military ethics. Balancing national security while preventing technological abuse and excessive concentration of power is a governance challenge faced by all leading technological nations.
Promoting two-way talent mobility is an effective path to bridging cognitive gaps. Alexander Karp and Nicholas Zamiska suggest establishing channels for technical experts to enter government departments while encouraging government officials to undertake temporary assignments at tech companies for learning purposes. This two-way flow helps break down information barriers, enabling tech companies to understand policy logic and governments to grasp the patterns of technological iteration. Obstacles to talent mobility are often harder to overcome than technological barriers.
Moreover, a hierarchical and categorized cooperation model can be established to avoid a 'one-size-fits-all' approach. Different technological fields and application scenarios require distinct government-enterprise cooperation mechanisms. Defense AI demands high confidentiality and strict regulation, medical AI necessitates data sharing and ethical review, while urban governance AI requires public participation and oversight. Attempting to address all scenarios with a single model will inevitably lead to collaboration failure.
The coupling between governments and enterprises in the AI era essentially represents a global renegotiation of technology governance. It encompasses multiple dimensions, including power distribution, responsibility delineation, and interest balancing. Technology companies need to recognize that AI is not a neutral tool but a complex system embedded within social relationships; governments, meanwhile, must update their governance frameworks to adapt to new challenges brought by rapid technological iteration.
For industry practitioners, regardless of the country or technological field they operate in, it is necessary to rethink several core questions: Where are the boundaries of technological innovation? How can commercial interests and public responsibilities be balanced? Between globalization and localization, how should technology companies position themselves?
From a global perspective, the long-term value of AI companies may increasingly depend on their ability to manage government-enterprise relations. This includes understanding policy trends, establishing compliance systems, participating in standard-setting, and fulfilling social responsibilities across multiple dimensions. The era driven purely by technology is passing, and the three-dimensional capabilities of technology, commerce, and policy will become the core competitiveness of AI companies.
Government-enterprise coupling is not a zero-sum game. Technology companies need governments to provide data, scenarios, and policy support; governments, on the other hand, rely on technology companies for innovation capabilities and solutions. The real challenge lies in finding autonomy within interdependence, maintaining independence in cooperation, and avoiding over-reliance in coupling.
▍Combatting 'Spiritual Hallowing,' Remodeling Corporate Beliefs with an Engineering Mindset
The technology sector is undergoing a profound value crisis. As the capital frenzy subsides, many companies find themselves left with only a collection of impressive data metrics but have lost their core purpose. This 'spiritual hallowing' manifests not only in financial statements but also permeates daily organizational operations. Meeting rooms are filled with meticulously polished PowerPoint presentations, yet few genuinely care whether products solve real-world problems. An engineering mindset offers an antidote to this sense of emptiness.
An engineering mindset first requires a clear definition of the objective function. In software development, this means defining precise inputs and outputs; in corporate management, it entails answering the question, 'What is our fundamental purpose for existence?' Traditional enterprises often prioritize profit maximization as the sole goal, but an engineering mindset suggests that excellent companies should resemble well-designed systems, seeking dynamic balance among multiple objectives.
Take SpaceX as an example: its objective function includes not only minimizing launch costs but also covers multi-dimensional indicators such as technological breakthroughs and ecosystem building. This multi-objective optimization mindset allows the company to strike a balance between commercial returns and technological ideals. A one-dimensional criterion for success often marks the beginning of spiritual hallowing.
The concept of the 'Minimum Viable Product' from agile development can be applied to organizational culture building. Many companies attempt to construct a perfect cultural system all at once, often failing due to excessive complexity. An engineering mindset suggests adopting an iterative approach, starting small, rapidly experimenting and correcting errors, and continuously improving.
$Netflix (NFLX.US)$ The cultural handbook serves as a typical example. This document, known as 'Freedom and Responsibility,' was not created overnight but went through dozens of iterations. Each version was optimized based on employee feedback and data validation, culminating in a unique management philosophy. The formation of culture should resemble software development—a continuous iterative process.
In the field of engineering, redundancy design is crucial for ensuring system reliability. Similarly, organizations need to maintain a certain degree of 'innovation redundancy' beyond their core operations. Google's famous '20% time' policy essentially reserves space for serendipitous discoveries. Although this policy has evolved over time, its essence continues to this day.
3M allows engineers to dedicate 15% of their work time to self-selected projects, a policy that has led to innovative products like Post-it Notes. Moderate resource redundancy provides fertile ground for innovation, whereas excessive pursuit of efficiency can stifle creativity.
Antifragility is another key concept in engineering thinking. In system design, antifragile systems become stronger under stress.$Amazon (AMZN.US)$ The 'two-way door' decision-making theory reflects this idea. The company categorizes decisions into two types: irreversible 'one-way door' decisions require careful analysis, while reversible 'two-way door' decisions encourage rapid experimentation. This classification reduces the cost of trial and error and fosters organizational antifragility.
Engineering thinking emphasizes first principles, which involve reasoning from the essence of things. In enterprise management, this means looking beyond industry conventions and traditional wisdom to directly consider user needs and physical limitations. When $Tesla (TSLA.US)$ Tesla designed electric vehicles, they did not follow the conventional path of the automotive industry but instead rethought battery chemistry and physics principles. This first-principles approach has brought about disruptive innovations.
In organizational management, it is equally important to return to first principles. Instead of imitating other companies' management practices, one should deeply consider: what kind of environment enables top talent to reach their full potential? What mechanisms ensure a balance between decision quality and speed?
The AIP Bootcamp exemplifies the raw beauty of engineering thinking. It is an extreme practice within Palantir: when faced with complex customer demands, the company forms a lean team to deliver a functional solution within five days. The entire process avoids sales pitches, PowerPoint presentations, or proof-of-concept demonstrations, diving straight into coding and system deployment.
In traditional sales culture, engineers are often required to learn 'customer communication skills' or 'value presentation methods.' However, Palantir takes the opposite approach: let those who understand technology best address the problem directly, solving issues with code rather than slides. If the system does not function after five days, the project is immediately terminated; if it succeeds, clients naturally buy in.
Behind this 'aesthetic of force' lies the engineering mindset's pursuit of ultimate efficiency. It assumes that most business problems can be quickly validated through technological means, while lengthy processes often serve as a protective layer for organizational inertia. When technical teams are empowered to take direct action rather than go through layers of reporting, the speed of innovation increases exponentially.
Though extreme, this practice reveals the essence of the engineering mindset: actions speak louder than words, and results outweigh promises. When companies become spiritually hollow, it is often because processes have replaced action, meetings have supplanted creativity, and reports have taken precedence over deliverables.
The principle of 'modular design' is equally applicable to organizational structures. Breaking down large organizations into relatively independent small teams can enhance flexibility and accelerate innovation. Haier’s 'Chain Group Organization' exemplifies this concept. The company organizes its employees into user-facing micro-organizations, each functioning as an independent operational unit. This modular structure allows this traditional manufacturing enterprise to maintain startup-like vitality.
In complex systems, engineers build dashboards to monitor key metrics. Similarly, enterprises need to establish their own 'organizational dashboard' to track truly important indicators, not just easily measurable numbers. Overemphasis on short-term metrics can lead to the trap of 'optimizing locally while sacrificing the whole.' Some e-commerce companies overly optimize click-through rates only to find that user loyalty declines; certain social platforms pursue maximum user time, resulting in deteriorating content quality.
Engineering thinking must ultimately serve broader systemic goals. Technology companies need to consider their role and function within the larger socioeconomic system instead of becoming trapped in inward-focused competition. When businesses use engineering thinking to reassess their mission, they can achieve commercial success while maintaining spiritual fulfillment and clarity of direction. This balance may well be the best antidote to 'spiritual hollowness.'
Editor/Rice