Molecules and Markets

A desk at 2am. An energy drink going flat. A cigarette, or something close to it. A laptop open, Claude finishing your code. Three stimulants, one goal: stay sharp long enough to get the work done. 

This blog plans to break down the chemistry behind caffeine and nicotine, then turns to the strangest stimulant of the two. One that does not just keep you awake. One that starts doing the work itself. 

Part 1: Caffeine 

Your brain has been keeping score all day. Every hour you stay awake, neurons produce adenosine, a byproduct of neural activity. Adenosine binds to A1 and A2A receptors throughout your brain. The more it accumulates, the stronger the signal it sends to slow down, rest and sleep. Caffeine is structurally similar enough to adenosine to fit into those same receptor sites. It binds to the adenosine receptor, but does not slow down cell activity the way adenosine would. The cell can no longer identify adenosine because caffeine is taking up all the receptors adenosine would normally bind to. Your brain stops receiving the fatigue signal. Without that brake, neurons fire faster. The pituitary gland senses this increased activity and releases hormones telling the adrenal glands to produce adrenaline. Your pupils dilate, heart rate climbs, airway opens slightly. Your body enters a mild stress response, not because something is wrong, but because you chemically convinced it something might be.

Here is what caffeine does not do: give you energy. The adenosine keeps building behind the blockade. When caffeine clears your receptors, usually four to six hours later, all that accumulated adenosine floods back in. With more receptors now available, the rebound feels even more intense once the caffeine wears off. The crash is not a side effect, it’s the debt.

Chronic caffeine intake leads to upregulation of adenosine receptors, an increase at the cellular level of the magnitude or rate of physiological response or biochemical process. Which reduces its efficacy over time. Your baseline shifts, more caffeine is needed to get the same block. You have built a chemical subscription you cannot cancel without a headache.

None of this stops you from drinking it. The caffeine market hit $495 billion in 2024. It is the only psychoactive drug with its own artisanal supply chain, its own vocabulary, and a cultural status so embedded that offering it to someone counts as hospitality. The productivity economy did not need to market caffeine. Caffeine did it itself.

Part 2: Cigarettes 

Nicotine has the worst branding in pharmacology. Your brain connects it to lung cancer, addiction, and early death. Those consequences belong mostly to combustion, not the molecule. Nicotine itself is one of the most precisely studied cognitive enhancers in neuroscience, and the data on it are clearer than most people know.

Contrary from the mechanism of caffeine. Rather than blocking a receptor, nicotine mimics acetylcholine, the brain’s own neurotransmitter for attention and learning. It binds to nicotinic acetylcholine receptors (nAChRs), particularly the alpha4beta2 and alpha7 subtypes, which are concentrated in the prefrontal cortex and hippocampus. Nicotine enhances attention and working memory by activating nAChRs in the prefrontal cortex, a region critical for these cognitive functions and rich in nAChR expression. Which is another reason why nicotine has been used to treat early stage Alzheimer’s and Parkinson’s disease.  

Research shows that in the short term, smoking “between Claude usage limits” or on breaks can improve attention, reaction time, working memory and attention, making it feel productive. The most significant action occurs in the prefrontal cortex, where the brain governs working memory, attention and decision-making. And also in the nucleus accumbens, which controls motivation and reward. The molecule also stimulates dopamine release from neurones in the ventral tegmental area, creating the reinforcement signal that makes the cognitive lift feel pleasant and repeatable. The combination of the improvement of focus and skills, and the “rewarding” effect, places smoking in the golden zone when it comes to the stimulant of choice for the younger working generation. 

Scientifically, the primary reason for continued smoking to stay focused is due to the difficulty concentrating and impaired attention that are core effects of nicotine withdrawal. In other words, the initial experience of smoking may have some benefits. However, following that, the main part of what a smoker experiences as the “benefits” of a cigarette between Claude prompts is the restoration of a baseline that nicotine addiction has itself eroded. The drug creates the deficit it appears to cure. 

However, the initial cognitive mechanism is real, and it has fueled an alternative market separated from tobacco. Nicotine pouches such as Zyns, which deliver nicotine through the gum without the need of combustion, have become more present in the exact environments where performance pressure is at its highest. Driven by both its measurable cognitive effects and its cultural framing, nicotine pouches have become a notable presence in places like trading floors, coding sessions, late night study rooms. These pouches market is growing at double digit tastes, particularly among young men (18-35). The product is not marketed as a drug but focus on sale. 

This is the same commercial reframing we identified in the Modafinil post, and again in the longevity industry covered in Tutto Passa: take a legitimate neurochemical mechanism, detach it from its most dangerous delivery system, package it as a lifestyle product, and sell it to people who would never describe themselves as drug users but will absolutely describe themselves as optimisers. In other words, these stimulants have now been embedded into the high paced work culture’s social handbook, not participating in these activities could make you an outcast, or “uncool”. 

Part 3: Claude

Arguably the most influential wave in the productivity economy or in the economy as a whole is AI. With the recent launch of Claude AI, the most advanced artificial intelligence, there is a cultural moment happening with startups and knowledge based work environments that deserves its own “molecular” analysis. In the same places where caffeine and nicotine have circulated, a third “cognitive” input has arrived: AI assistance, used centrally to streamline and cut down on “grunt” work. It serves as a real time extension of working memory and analytical capacity. 

Especially in the startup culture, the combination of stimulants and AI, aiding less programming experienced young individuals to start programming based businesses has taken its own name, “vibe coding”. The process of building software not by writing every piece of code but by describing what you want, reviewing what the model produces, adjusting direction and correcting errors, often using ChatGPT in combination to correct errors. The individual provides judgement and intention with “feel”, while the model provides the expertise and speed. The combination in right hands has allowed a number of start ups to reach levels that previously would have needed an army of workers. 

The numbers from Anthropic’s own research tell this story clearly. In January 2025, 36% of jobs in one sample saw Claude being used for at least a quarter of their tasks; pooling data across reports, this has risen to 49%. Computer and mathematical tasks dominate usage accounting for about a third of all conversations on Claude.ai and nearly half of API traffic.

Most significantly, Antrophic’s latest report on the concept of “observed exposure”, a measure of which professional tasks AI is actually performing, not just theoretically capable of performing. Theoretical AI coverage exceeds 80% in several industries such as computer and math as well as business and finance, with the highest theoretical coverage at 94.3%. But actual observed use is currently a fraction of that. In computer and mathematical occupations, AI systems could theoretically handle 94% of tasks, but actual AI usage currently covers around 33%. 

The gap between what  AI can theoretically do and what it’s currently doing is the most important economic number which no one is talking about. As capabilities advance and adoption deepens, the “red area” of acquittal use will grow, filling the “blue area”  of what is possible. When it does, the disruption to white-collar knowledge work will be structural not incremental. 

Anthropic CEO, Dario Amodei said, the technology could disrupt half of entry-level white-collar work. The economic index data suggest this is not a distant prediction. It is an early-stage process already underway, concentrated so far among younger workers, where there is suggestive evidence that hiring of younger workers has slowed in exposed occupations.

However, the leading AI chip maker Jensen Huang of Nvidia offers a different perspective. While AI can match or outperform humans in specific tasks, such as detecting cancerous tumors in radiology, this has not reduced demand for radiologists. In fact, the opposite trend can be seen. 

The reason is structural. AI improves detection rates, which leads to more diagnoses. More diagnoses lead to more patients entering treatment pipelines. This increases healthcare activity and revenue which in turn raises demand for radiologists rather than replacing them. Radiologists do not only identify tumors, they interpret results, make clinical judgements, and coordinate patient care alongside other doctors. 

A similar pattern can be seen within Nvidia. Despite the rise in AI, Huang argues that human input remains essential, and the number of software engineers is expected to increase, not decline. AI can process information and identify patterns, but it cannot independently define problems, connect disparate ideas or lead complex decision making processes. 

This points out a clear implication. The competitive advantage is shifting towards skills AI cannot replicate such as judgement, synthesis of ideas and leadership. To remain relevant, individuals must specialise in these higher order capabilities rather than routine analytical tasks. 

Finally, what connects caffeine, nicotine and Claude is not just that they are all used to enhance cognitive performance. It is that each one reveals something about the economic value placed on sustained human attention. 

Caffeine demoralized wakefulness during the industrial revolution and the rise of office work. Nicotine became the performance drug of high stakes professional environments, trading, finance, law and politics. And now AI assistants are doing something subtler and more structurally significant, they are not just extending human cognitive capacity, but are partially replacing the need for it in an expanding set of tasks. Caffeine blocks the signal that tells you you’re tired. Nicotine extends the window of focused attention. Claude handles an increasing share of what that attention was supposed to produce.

The coder working at 2 am still relies on caffeine, just as the industrial worker once did. The same dependence on nicotine persists, from office workers to finance executives. The habits of productivity have not changed, the scale has. 

What once took entire teams, now with the help of Claude is achievable with a single individual. The combination of chemical stimulation and AI assistance has reshaped how individuals work, think and produce. This pairing amplifies focus, extends working capacity and accelerates output. Productivity is no longer just about effort, it’s about leverage. Today this leverage comes from the intersection of chemicals, habit and AI.