Something is happening. No big moment. No tipping point. A shortcut here, a suggestion there. Quietly, steadily, AI has started doing the thinking for us. And the more it does, the less we seem to do.

This isn’t about the decline of intelligence (although research suggests that’s happening as well). It’s about the decline of effort, about process, about how we learn and make decisions. And that’s what’s starting to change, especially for the next generation.

There’s a term for this shift: cognitive offloading. We’ve been doing it for years. Calculators take care of maths. Satnav gives us directions. Google answers every argument. Now AI writes our emails, summarises meetings, answers homework, brainstorms ideas, builds pitch decks, and even composes love letters.

None of this is bad in itself, but when it becomes default behaviour - when we skip the struggle every time - we start to lose the skills we used to rely on. Ask AI to do the thinking often enough, and eventually, you forget how to do it yourself.

If my teacher friends are right, you’re starting to see this most clearly in schools. It’s not just that students are using AI to cheat on essays; it’s that they’re using it to avoid the slow, messy, necessary work of figuring things out. They type in a question, copy the answer, and move on. It’s unquestionably efficient, but it’s also skipped the part that matters… where learning actually happens - the bit where you get stuck, where you question, where you try again and, eventually, start to understand.

And it’s more than just the answers being outsourced. It’s also the curiosity and the critical thinking. Kids are becoming prompt engineers, not problem solvers. They’re getting used to answers without ever needing to ask why. A recent FT article reports that problem-solving confidence among young adults is lower than ever. One Gen Z respondent said they feel mentally weaker than their parents—not because they know less, but because they’ve never had to sit with a problem and figure it out.

It’s not just anecdotal. OECD data shows adults are struggling too. Across developed countries, basic reasoning and comprehension are down. At work, the pattern is the same. It’s not just admin that’s being automated; it’s the thinking too. Research shows that the more people trust AI, the less they check its output. This isn’t efficiency; it’s complacency.

At home, it’s no different. AI curates your news feed, your shopping list, your watch list. Recommendation engines decide what you see. Algorithms get so good at guessing your preferences that you stop exploring anything else. It becomes a kind of mental autopilot. Why read the book when TikTok already summarised it? Why choose what to watch when Netflix has something ready? Why plan dinner when your fridge already has a suggestion?

We’re also reading less. We retain less of what we read, and we struggle to analyse it critically. Our attention is spread thin across platforms designed to keep us scrolling, not thinking. Long-form content feels harder to engage with. Even when we click on an article, we often skim the headline, glance at the pull quotes, maybe dip into the comments—but rarely the full piece. Viral content grabs us by the feelings, not the facts. Context is stripped away, nuance gets lost, and we’re more likely to share than to sit with an idea. It’s not that we’ve lost the ability to read deeply. We’ve just fallen out of the habit. And like any muscle, leave it unused long enough, and it starts to atrophy.

And when those habits go, so does our judgment. Without deep reading, there’s no deep thinking. Without thinking, there’s no discernment. And without that, we’re just drifting, scrolling through a world increasingly shaped by machines we don’t fully understand and content we no longer question.

This isn’t a call to reject AI. Used well, it’s a wonderful tool. But like any tool, it depends on the hand that holds it. If we’re not paying attention, we risk trading too much; not just time and effort, but the mental grit that helps us grow sharper, more curious, more capable.

So maybe the challenge isn’t whether we use AI; it’s whether we still think after we do. Whether we ask the second question. Whether we rewrite the first draft. Whether we keep some struggle in the system. Think of it like power steering. Great if you’re awake. Useless if you’ve nodded off.

That’s the risk. Not the machines rising up. But us switching off.

They say one man’s trash is another man’s treasure. This quote becomes very literal when you consider that preindustrial workers built entire industries based on human urine—it was so valuable that the ancient Romans not only sold pee collected from public urinals, but those who traded in urine had to pay a tax.

Prior to the ability to synthesise chemicals, urine was a quick and rich source of urea, a nitrogen-based organic compound. When stored for long periods of time, urea decays into ammonia.

Ammonia has a multitude of properties—when added to water, it’s a caustic that breaks down organic material, making it the perfect substance for softening and tanning animal hides. As a base, ammonia is a useful cleanser because dirt and grease—which are slightly acidic—get neutralised. Ancient launderers would use urine as a soaking treatment to get tough stains out of cloth.

Ammonia was also a mordant (a dye fixative), helping to stop natural dyes from leaching by fixing colors to cloth. Urine was so important to England’s textile industry in the 16th century that casks of it were shipped from London to Yorkshire. The production rate required about 200 tonnes of urine per year, the pee of at least 1,000 people.

You’ll be heartened to hear that a French startup, Toopi Organics, is very much keeping the tradition alive. They collect urine and put it in biofermenters, where it becomes a growth medium for microorganisms. The finished product is a bacteria-concentrated liquid that can be used as an effective and environmentally friendly biostimulant for agricultural use.

Toopi is one of many extraordinary companies enabling waste valorisation—the process of reusing, recycling, or composting waste materials and converting them into more useful products, including materials, chemicals, fuels, or other sources of energy.

It might sound like the stuff of science fiction, but thanks to some maverick inventors and an influx of investment, we can now turn air into aviation fuel, corn into bleach, and even cow pats into graphene.

Take Air Company, they’re decarbonising the planet by transforming CO2 into an endless resource, converting carbon into carbon-negative chemicals and fuel. The company is developing technology (dubbed AIRMADE™) that uses (primarily) biogenic CO2 and green hydrogen to produce alcohols and paraffin that are of fuel-grade quality.

Meanwhile, Levidian (another decarbonisation company) has found a way to literally “polish a turd.” Their LOOP technology is a self-contained modular system that can take biomethane (produced from cow slurry) and transform it into clean-burning hydrogen and wonder material graphene.

Solugen have slightly larger ambitions, they’re trying to transform the entire petrochemical industry using biology. They’ve raised over $600 million in venture funding and now have a scaled chemical plant in Houston that uses a “chemienzymatic” approach to turn feedstocks such as dextrose (corn sugar) into tens of thousands of liters of bio-based chemicals, including hydrogen peroxide (bleach), gluconic acid, and glucaric acid.

While Solugen uses sugar as their input, Supplant creates sugar as their output. They’ve developed a biotechnology platform that takes plant waste and uses a single enzyme reaction to produce a cane sugar alternative that has the same physical properties as regular sugar and can be used as a like-for-like substitute in baked goods. As an added bonus, the body processes Supplant Sugar as if it were fibre, so it’s healthy.

And then there’s Pipeline Organics, who use wastewater as a feedstock to produce clean, cheap bioenergy. They’ve created a high-efficiency enzymatic biofuel cell that generates electricity through the breakdown of liquid organic waste during wastewater treatment.

These world-saving companies have only been made possible by scientific advances and extraordinary entrepreneurship. We also have to thank the rising tide of venture capital taking the plunge and investing in circularity—after all, these breakthroughs are not cheap and most will never leave the lab.

This increase in awareness and investment is no doubt buoyed by two geopolitical issues: climate change and national security.

The climate crisis demands a transition to sustainable practices. Both governments and consumers are becoming increasingly climate conscious and showing a growing interest in paying a “green premium” for eco-friendly products and processes.

At the same time, deglobalisation and national security concerns are prompting countries to embrace self-reliance. It’s becoming increasingly important to localise supply chains and maximise the value of domestic resources.

In light of these developments, it’s clear that humanity is on the cusp of a new industrial revolution—one that prioritises sustainability, innovation, and the efficient use of resources. These pioneering companies are not only mitigating the impact of climate change but are also fostering a more self-sufficient and resilient global economy.

The shift towards waste valorisation and the creation of the precious from the mundane is testament to the power of technology and the promise it holds. As we continue to push the boundaries of what’s possible, we must support and invest in these groundbreaking ideas and the visionary entrepreneurs behind them.

As they say, necessity is the mother of all invention.

Till next month.

In 1911, the US Supreme Court ruled that Standard Oil - then the largest oil producer in the world - was an illegal monopoly.

It was a landmark decision that resulted in the dissolution of Standard Oil into 34 smaller companies. It also made John D. Rockefeller the richest person in modern history and laid the foundation for modern antitrust authority in the United States, serving as a blueprint for the rest of the world.

Now, trustbusters exist in every developed nation. Their raison d'être is to protect consumers by punishing cartels, policing mergers, and dealing with dominant firms that abuse their positions. In America, it’s the Federal Trade Commission (FTC) and the Department of Justice (DOJ). In Europe, it’s the European Commission and national regulators (i.e., Germany’s Bundeskartellamt), and in the UK, it’s the Competition and Markets Authority (CMA).

113 years after Standard Oil, there's a new titan for regulators to contend with: "Big Tech." But this time, it’s not about oil; it's about AI, and the tech giants are tearing up the rulebook.

For years, "Big Tech" has stayed one step ahead of the competition authorities. This dynamic has been characterised by a cycle of innovation and adaptation, where the tech giants continuously push the boundaries of technology, data usage, and market expansion, while regulators scramble to understand these advancements and craft appropriate responses.

The complexity and rapid evolution of technology pose significant challenges for regulation. While the world advances, the watchdogs are anchored in the past. Much of the body of law applies to old industries and must be adapted to present circumstances. Prosecutors are struggling to fit the modern world into the doctrines of old law.

The tech giants have exploited these regulatory gaps and ambiguities and mastered the art of stalling and obfuscation. By the time regulators catch up to one issue, new technologies and business models have already changed the landscape, starting the cycle anew.

Moreover, "Big Tech" has amassed significant resources and influence, which they can deploy to shape regulatory environments to their advantage. Through lobbying, legal challenges, and public relations campaigns, these companies influence policy and public opinion, further complicating regulatory efforts.

But the real shenanigans have only just begun - as the tech giants desperately rush to secure AI market dominance - all kinds of extraordinary manoeuvres are starting to take shape.

Last month, the Competition and Markets Authority shared concerns that established companies are taking advantage of AI companies’ insatiable need for computing power to train large language models by luring cash-strapped startups to their cloud services in exchange for a stake that could give them outsized influence.

This is exactly what is happening, and as per usual, the regulators are just a little slow to catch on. Six months ago, I wrote about the concept of ‘round tripping’ and a startup called CloudWeave - and how they were using Nvidia’s much-sought-after H100 AI GPUs as collateral to raise debt to buy even more GPUs from (you guessed it) Nvidia …

And then there’s Amazon putting $1.25 billion into AI startup Anthropic in exchange for a minority stake, and certain tit-for-tat agreements (like the company continuing to use AWS for its extensive computation needs). A few months later, Amazon ponied up another $2.75 billion on the understanding that Anthropic would build its AI using specialised computer chips designed by (you guessed it) Amazon …

While everyone is playing checkers, Microsoft is playing chess. Their machiavellian genius was first evident back in November ‘23 when OpenAI shocked the tech industry with the surprise ousting of its CEO - Sam Altman.

Four days later, the decision was reversed. Microsoft had used its considerable leverage - including some of the $13 billion it had pledged but not handed over - to pressure the board of OpenAI to reverse the firing of its CEO, get him reinstated, and then appoint an almost entirely refreshed board with Microsoft as a member (albeit one without a formal vote).

Despite their 49% stake, a board seat, and obvious executive power, Microsoft apparently neither owns nor controls OpenAI. Instead, they take great pains to talk about their investment and “partnership”. It’s what you might call “heavily lawyered” – that is to say, carefully constructed to avoid looking like a buyout. But when the chips were down for Sam Altman, in the end, Microsoft was holding all the cards – not OpenAI’s board.

In June ‘23, Inflection - a much-vaunted AI startup co-founded by Reid Hoffmann, Mustafa Suleyman, and Karén Simonyan - announced it had raised $1.3 billion to build what it called “more personal AI.” The lead investor was Microsoft.

Less than a year later and out of the blue, Microsoft announced that it was absconding with the co-founders of Inflection, much of the staff, and the rights to use the tech. Apparently, it paid $620 million for non-exclusive licensing fees for the technology and $30 million for Inflection to agree not to sue over Microsoft’s poaching.

I’m sure it was an impossibly enticing offer for Mustafa, Karen et al. No doubt compensation packages featured salaries and stock in the tens of millions, all of the compute they could possibly want, and the opportunity to work alongside some of the brightest minds in AI. It’s a talent snatch without the acquisition - the ultimate power move - a bloodless coup that gets Microsoft top tech talent and leaves regulators scratching their heads.

You can’t help but feel sorry for the competition watchdogs; they’re bringing bows and arrows to a gunfight!

Till next month.

As even casual pop-culture fans know, The Matrix is set in a distant future where machines have gained sentience and conquered humanity. Humans are stored in vats and used as batteries (the machines feed off their bioelectric and thermal energy). To keep humans dormant and placid, their consciousness is inserted into ‘The Matrix’ a simulation engineered by the machines - a virtual reality mirroring life on earth in 1999.

At the time, the films buzz was largely due to its groundbreaking special effects, gravity-defying stunts and cyber-punk aesthetic. Its sticking power is largely due to its core techno-philosophical question that has become the staple of every stoner conversation: Are we living in a simulation?

“You take the blue pill, the story ends, you wake up in your bed and believe whatever you want to believe. You take the red pill, you stay in Wonderland, and I show you how deep the rabbit hole goes.” - Morpheus (The Matrix)

This question has a name: the simulation hypothesis. Proposed by the philosopher Nick Bostrom in 2003, the concept suggests that our reality might actually be a computer simulation created by a highly advanced civilisation.

1. Advanced Technology: Bostrom suggests that if a technologically advanced civilisation exists with the capability to run incredibly complex simulations, they might create simulated universes for various reasons, such as entertainment, research, or even to understand their own history.

2. Many Simulations: If such a civilisation exists and has the ability to create simulations, they might create many of them, possibly even more simulations than there are real, "base" realities.

3. Simulation Realism: Bostrom argues that if these simulations are realistic enough, the inhabitants within them (like us) would not be able to distinguish between their simulated reality and a "real" one. To them, their world would seem just as real as ours seems to us.

4. Implications: If we accept Bostrom's hypothesis as a possibility, it raises profound questions about the nature of reality and our place within it. It also suggests that our understanding of the universe might be limited, as we could just be characters in someone else's simulation.

In short, the closer technology gets to being able to build a fully interactive simulation like the Matrix, the greater the possibility that someone has built such a world and that we’re living inside it.

Further, if you assume that a “technologically mature” civilisation can easily create a simulated world, then logic suggests that said civilisation should be able to to create multiple worlds, each with a multitude of simulated characters.

With these simulated worlds far outnumbering our “real” world, perhaps into the billions, the likelihood that we are in a simulation would be significantly higher than not. It was this logic that prompted Elon Musk to suggest, a few years ago, that ‘there is a one in a billion chance that our world is a base reality’. A more recent study suggests chances are a more like 50/50.

It’s a theory that is difficult to prove — but difficult to disprove as well. Bear in mind that an AI-generated world would be indistinguishable from our physical world. Either this ‘reality’ is being streamed directly into your brain, or we are simply AI-generated characters living inside a simulation.

Decades later and the simulation hypothesis is starting to be taken a lot more seriously by technologists and scientists. This is largely due to recent technological advancements in artificial intelligence, brain computer interfaces, computer graphics, processing power and augmented reality / virtual reality.

Consider the progress from just this year:

• Apple released their highly anticipated Vision Pro headset - a mixed reality device that is a huge leap for spatial computing and the merging of digital and physical worlds. 

• OpenAI unveiled Sora AI, a generative AI that can create realistic and imaginative scenes from text instruction that is virtually indistinguishable from real captured footage. 

• Neuralink successfully implanted their first chip into a human brain. As a result, Noland Arbaugh, a 29 year old quadriplegic can now transmit data wirelessly and move a computer cursor using just the power of his mind.

• Epic Games dropped Unreal Engine 5.2, their most advanced real-time 3D creation engine that can generate nearly photo-realistic procedurally generated gaming environments. 

• Nvidia announced their Blackwell GPU - the worlds most powerful chip will power the next generation of AI - it represents a 1000x increase of compute power compared to chips produced 8 years ago. 

Individually these technologies represent stunning achievements, collectively they suggest the simulation point, a sort of singularity when AI-generated virtual reality becomes indistinguishable from physical reality, is imminent.

Even if you discount the notion that we are already part of a simulation - there’s plenty to suggest we’re heading in that direction. Many of us are already servants of social media, which has profound implications for how we perceive the world.

This digital bubble, driven by algorithms and personalisation, acts as an echo chamber where our beliefs and opinions are constantly reinforced, our perspective is never challenged and alternative viewpoints are virtually non-existent. Consequently we view the world not as it is, but as we wish it to be.

While companies like Meta are busy building the metaverse. Over half the internet is already AI generated content. Millions of us chat regularly with AI characters, form relationships with avatars, socialise in virtual worlds and grew up on multiplayer online games like Fortnite and Roblox.

Whether we like it or not, much of our world is filtered through the prism of digital platforms. What we know, what we watch, what we learn, how we live, how we socialise – all of these modern human experiences are already influenced by algorithms that direct us in subtle but serious ways. As Philip K. Dick, the seminal sci-fi author put it, all the way back in 1977, “We are living in a computer programmed reality.”

Till next month.

Maybe I’m idealistic, but I’ve always imagined a world where AI liberates us from the tedium of work, freeing up our time to pursue human expression and the higher arts. I’ve assumed that creatives would be insulated - that they would be the last in the firing line. Now, I’m not so sure.

A few weeks ago, I was lucky enough to attend a pre-screening of Dune: Part Two. I’m no critic, but I can confirm that the film is absolutely epic. The director (Denis Villeneuve) is known for making visually inventive, sensitive, and unflinching films. It’s very clear with this sci-fi blockbuster, that he’s at the top of his game.

The film is a faithful adaptation of the self-titled novel “Dune” by Frank Herbert, first published in 1965, a science fiction classic that transports the reader to the fictional desert planet of Arrakis (aka Dune) tens of thousands of years into the future.

It’s a multi-layered work that explores themes of politics, religion, ecology and the human experience. But there’s a cautionary sub-narrative that is largely missing from the films. The Orange Catholic Bible (OCB), the key religious text in the Dune universe, forbids the creation of machines that imitate human thinking: "Thou shalt not make a machine in the likeness of a man's mind."

There are no computers, robots, or AI in Dune because their universe has already gone through the full cycle of man vs machine - an age-long intergalactic war between humans and artificial intelligence “the god of machine logic” that ends in the total destruction (and ongoing ban) of thinking machines.

So, as I sat marvelling at the film - the culmination of so much vision and talent - I couldn’t help but think about Frank Herbert’s foresight, the inexorable creep of AI, and the uncertain future it poses for creatives.

That very same day, OpenAI, the US-based AI research organisation, had just announced ‘Sora’ - a generative AI model that can create realistic and imaginative scenes from text instructions. Not only does the model understand what the user has asked for in the prompt, but also how those things exist in the physical world.

Experts have hailed Sora as a major milestone in the development of artificial intelligence. It has caused widespread astonishment (and quite a lot of anxiety) with its ability to produce up to a minute of realistic footage from simple text prompts.

To put things into perspective … here’s a (now viral) video generated by open source "text2video" ModelScope AI in March 2023.

And here’s a video generated by Sora less than a year later.

Sora is just the latest example of many Generative AI tools that are taking on the creative category; text to image (ie. Midjourney), text to music (ie. MusicML), text to talking, singing, speaking (ie. Emote), text to games (ie. GenieAI)… the list goes on.

The questions posed by Generative AI are consequential (if not existential) for creatives. These tools have put the power of creation in the hands of anyone able to type a description. On the one hand, this can only be a good thing - equal, affordable opportunity for everyone. Just imagine the kind of talent that might emerge if a kid has the same access to visual effects as a Hollywood studio?

On the other hand, and as evidence is already showing, this democratisation comes hand-in-hand with the enshittification of absolutely everything. I know there’s no such thing as an original idea — that everything is a copy, but with so much mediocre content, it’s really starting to feel like everything is derivative. 

AI junk is already clogging the internet - filling it up with “zombie content” designed to game algorithms and scam humans. It's becoming a place where bots talk to bots, and search engines crawl a lonely expanse of pages written by AI.

Amazon is flooded by shitty AI-generated books. Twitter has a bot problem, Linkedin is a minefield of pointless posts and you can’t move for fake news - NewsGuard has so far identified 725 AI-generated news and information sites operating with little to no human oversight.

“Even a man's exact imitation of the song of the nightingale displeases us when we discover that it is a mimicry, and not the nightingale.” — Immanuel Kant.

What gives me comfort is that even with open access to all these tools, I still believe creativity will reside with creatives. Human expression is hard work! The tough part of creating content is crafting something from nothing. The truth is that most people don’t know what they want, and definitely can’t actualise their desires.

It is extremely difficult to create something from scratch or even just articulate an idea compared to picking among choices. It's hard enough to decide what to watch on Netflix! I have faith that most people will prefer to take the less energy-intensive route and leave creativity to the professionals.

I just hope that creatives will continue to strive for their art - I have no doubt it’s extremely tempting to let AI finish projects or use it as a muse - I had a conclusion in mind, but nothing encapsulates my feelings better than this snippet of a letter written by the indomitable Nick Cave, Australian musician, author and fearsome critic of Generative AI:

That ‘songwriter’ you were talking to, Leon, who is using ChatGPT to write ‘his’ lyrics because it is ‘faster and easier ,’is participating in this erosion of the world’s soul and the spirit of humanity itself and, to put it politely, should fucking desist if he wants to continue calling himself a songwriter.”

I don’t know about you, but there’s something quite unnerving about Figure 01, the eponymous autonomous humanoid creation of robotics startup Figure. Maybe it’s the blank iPhone-like face, maybe it’s the form-fitted wetsuit. Maybe it’s the sense of latent menace.

This feeling has a name. In the 1970’s, Masahiro Mori, then a professor at the Tokyo Institute of Technology, observed that as robots appear more humanlike, some observers' emotional response to the robot becomes increasingly positive and empathetic, until it reaches a point beyond which the response quickly becomes strong revulsion. He called this bukimi no tani, literally translated as ‘uncanny valley’.

Well, the uncanny valley is coming to a manufacturing facility near you. Earlier this month BMW, announced a ‘first-of-its-kind’ deal with Figure to deploy their autonomous humanoids at one of the car-maker’s factories in Spartanburg, South Carolina.

Robots in factories aren’t anything new, but they’ve mostly been single-purposed machines only capable of performing specific preset tasks. Humanoid robots promise a new level of adaptability and versatility.

Unlike their single-function counterparts, humanoid robots are designed to mimic human capabilities, featuring advanced hands, arms, and legs that enable them to handle a variety of tasks with agility and precision. They are the same shape as us, they can access the same spaces and use the same tools.

Digit from Agility Robotics has nimble limbs and a torso packed with sensors that enable it navigate complex environments and be used as a "mobile manipulator" carrying bins and empty totes around in warehouses where there's not enough space for conveyor belts.

EVE from 1X comes with strong grippers for hands, cameras that support panoramic vision and two wheels for mobility. EVE’s are reportedly already working as security guards at a couple of industrial sites in Europe and the US.

Phoenix from Sanctuary has proprietary haptic technology and human-like hands and arms which gives it dexterity to complete tasks that range from stocking shelves and unloading trucks to running registers.

These early models might move like marionettes and have the agility of geriatrics, but it’s getting easier to imagine a future where robots work side-by-side with their human counterparts, taking on dangerous, tiresome, repetitive tasks and freeing us from the burden of physical work.

That future is still a long way away, despite the BMW-buzz. To get to the point where robots are equal to or greater than their human counterparts, engineers need to solve some of the following, not inconsiderable challenges:

1. Actuators - these are machine components that achieve physical movements by converting energy into mechanical force. They are effectively the muscles of the machine. Future actuators need to have exceptional dynamic range - able to switch in a moment from powerful, rapid movements, such as lifting weights, to delicate, precise actions, like threading a needle.

2. Energy - powerful actuators will necessitate robust energy systems. These batteries will need to store and deliver sustained power over extended periods while also accommodating rapid and intense bursts of energy when needed. A high-performing robot should operate for extended durations before requiring recharging or will have batteries that allow for rapid charging cycles to ensure continuous functionality.

3. Sensors - to emulate human capabilities, robots will need a comprehensive sensor suite encompassing all human senses. This includes vision, hearing, smell, balance, proprioception (sense of how ones body, limbs and muscles are orientated), touch for grasping and collision detection, and even a sense of pain to prevent damage.

4. Compute - the "brain" necessitates powerful processors, capable of complex tasks, such as image recognition, natural language processing, and spatial awareness. Moreover, the computing architecture must strike a balance between processing speed and energy efficiency to ensure optimal performance without compromising the overall energy budget of the robot.

5. Control - to navigate the complexities of the world, a robot has to be capable of real-time decision-making, learning from experiences, and adapting to dynamic surroundings. It needs to take inputs from its sensors, make a model of its environment and understand how to orientate itself. It needs to determine its objective and make a decision about how to achieve that objective. It needs to plan its movements inline with its surroundings. It needs to instruct motor controllers to move its limbs to locomote and engage with objects in its space. This measure-plan-move sequence must also continuously adapt as the robot moves through its environment and as people and objects move around it.

To solve these challenges will require considerable breakthroughs in AI, digital electronics, material science and mechanical engineering. Just the energy issue may necessitate an entirely new source?!

One out-there idea is to use biobatteries (energy storage devices powered by organic compounds). The robot would have enzymatic components allowing it to “eat / drink” organic material which would then be converted into energy via enzymes; very similar to how we break down food. Another is to use a radioisotope thermoelectric generator (RTG), a type of nuclear battery that converts the heat released by the decay of a radioactive material into energy.

Some advances are happening faster than others. Last year, Toyota Research Institute (TRI) and Columbia University announced the development of Diffusion Policy a new, powerful generative-AI approach that enables easy and rapid behaviour learning from demonstration training.

It’s a big step. Previous state-of-the-art techniques to teach robots new behaviours were slow, inconsistent, inefficient, and often limited to narrowly defined tasks performed in highly constrained environments. Roboticists needed to spend many hours writing sophisticated code and/or using numerous trial and error cycles to program behaviours.

With Diffusion Policy, robots can now watch, learn and replicate what they see. They can be taught more human-like routines and motions without the typical complexity and expense. As an example, Figure (the aforementioned robotics startup) recently demoed Figure 01 making coffee by putting a capsule in a coffee maker.

Whilst this might not look particularly impressive - it’s worth noting that the robots end-to-end AI system was trained by watching humans make coffee, in just 10 hours. Not only could it insert the capsule and get the coffee machine started, it also learned to self correct mistakes. To make matters more impressive, this autonomous action is now transferrable to any other Figure robot running on the same system via swarm learning.

These skills are not limited to just “‘pick and place” or simply pushing around objects. Thanks to the groundwork laid by Diffusion Policy robots can now interact with the world in varied and rich ways. There’s still a huge amount of work still to be done, but its a significant step to ultimately building “Large Behaviour Models (LBMs)” for robots - analogous to the Large Language Models (LLMs) that have revolutionised conversational AI.

Till next month.

In 2008, in the midst of the financial crisis, the Fed cut interest rates as part of a frantic effort to avert a second Great Depression. The Fed also enacted quantitative easing, buying securities itself in an effort to push effective interest rates below zero. This "zero interest-rate policy" - otherwise known as ZIRP - was only meant to be temporary - but, when Congress wasn't forthcoming with stimulus spending, the Fed decided to keep ZIRP in place indefinitely.

With rates so low, investors were no longer able to earn nice returns by stashing money in safe assets - like US Treasury bills. Borrowing costs plummeted, cheap money proliferated and investor capital flowed into riskier asset classes - like publicly traded stocks or property development.

This influx of investment sent asset prices soaring. In what’s become known as '“the everything bubble”, almost every single asset class, from bonds, to private equity, to cryptocurrency sky-rocketed. The US stock market rose more than 580% after the financial crisis, accounting for price gains and dividend payments. At the same time, the tech giants Amazon, Netflix and Tesla set a new precedent for how valuable a technology company could become.

The Global Financial Crisis kick-started the era of easy money, and venture capital (VC) saw a surge in popularity, with many individuals entering the field due to high returns and a perceptions of it being easy and fun. As one of the few legitimate financial products with the potential to offer a thousand-fold return on investment, the VC ecosystem became absolutely flush with cash - and they needed to spend it.

Treasury investors shifted to corporate debt. Public equity hedge funds shifted to late-stage private equity. Late-stage private equity shifted to mid-stage, mid-stage to early stage. Seed rounds become bigger. Angel investors become a thing. Unicorns, unicorns, and more unicorns. [ReadMargins]

By 2015, global VC funding amounted to £128.5B. By 2019, it was $254B. Most of this money flowed into tech startups, from 2010 through 2020, tech startups made up a majority of venture funding. VC loves tech startups, they typically require small amounts of capital to start and are relatively capital-efficient to scale. With low/no marginal costs, many can benefit from outsized returns to effort ie. WhatsApp: 55 employees serving 420 million users and selling to Facebook for $19B.

In the face of so much capital, companies succumbed to temptation - soon enough everyone was a ‘technology company’ raising at tech valuations under the guise of innovation and and disruption. From Casper, an online mattress company ($100m raised at $1.1Bn valuation), to Zume, a pizza delivery company ($375m raised at a $1.5Bn valuation), to Juicero, a fruit and vegetable juice company ($120m raised at a $400m valuation).

We’ve all come to use the phrase “ZIRP phenomenon” to describe inflated VC rounds, inflated salaries, generous workplace perks, gravity-defying valuations, and logic-defying business models. [Angular Ventures]

Plentiful money fuelled exuberance — founders were encouraged by their investors to spend money like it was going out of fashion. ‘Growth at all costs’ became an entrenched part of the Silicon Valley zeitgeist. Tech companies began to ‘pay to win’ spending heavily on advertising, sales and user acquisition, buying users and growth (and the high valuations that comes with it).

Credit to Tony Yiu for a beautiful summary of this unsustainable cycle:

1. Venture capital rushes into startups.

2. Those startups spend that money on marketing (and consuming each other’s SAAS software), which allows them to rapidly scale their user bases and revenues (but no profit because spend > sales).

3. The rapid growth pushes up valuations of those startups, which attracts more venture capital at even higher valuations. Naturally, this also allows the venture funds that got in previously to record very high returns.

4. The new capital from step 3 lets the startups spend even more on user acquisition, advertising, etc., continuing the cycle.

The era of easy money snowballed exponentially until a COVID crescendo. In March 2020, as part of wide-ranging emergency action in the face of the coronavirus pandemic, the Fed repeated history, dropping interest rates to ~zero, restarting quantitative easing and distributing trillions of dollars to large swathes of the US population as part of lockdown relief.

Cashed up, locked in and with little to do in their leisure time, consumers moved online and eCommerce exploded. Americans spent $1.7 trillion online during the last two years of pandemic. That’s $609B more than the two years before Covid. The pandemic also accelerated digital transformation. Businesses and individuals increasingly relied on technology to adapt to remote work, online education, telehealth, and virtual communication.

The surge in remote work and reliance on digital platforms led to increased demand for technology services. Companies providing collaboration tools, cloud computing, cybersecurity, and e-commerce solutions, among others, experienced heightened demand for their products and services

Technology was having its heyday and everyone piled in. Hedge funds, private equity firms, sovereign wealth funds, corporate VCs and mutual funds supplied two-thirds of the $621B invested into venture capital 2021. The size of venture funds exploded as investors put ever-larger amounts of capital to work. As demand for deals grew, so did valuations. In a seller’s market, founders had the driving seat when it came to valuations and terms.

Fear of missing out (FOMO) proliferated and it wasn’t just high prices investors were prepared to pay to not miss the boat: investment discipline loosened. The periods for conducting due-diligence drastically shortened, protective provisions disappeared from term sheets and investors no-longer demanded board seats. At one point “crossover” fund Tiger Global Management was doing a deal a day, including on weekends.

Stupid ideas and shady founders enjoyed easy access to hundreds of millions even billions of dollars. Unicorns proliferated - more unicorns were minted in 2021 than the past five years combined and silly season began. Hopin, a virtual events company, hit a $5B valuation within 21 months of launch. Gorillas, an on-demand delivery company, achieved a >$1B valuation within 9 months. Cazoo, the online car retailer, surpassed a $1B valuation within 6 months.

With massive multiples, revenue growth driven in part by government stimulus, money drops, and low cost of capital, tech companies went on hiring rampages. Tech wage inflation accelerated boosting compensation for key roles by up to 20%. Startups, often pre-product market fit but buoyed by venture capital, frequently hired rapidly ahead of revenue and traction.

In the past a $1M ARR company would often have 5-20 employees. During COVID, some companies ballooned their employee counts regardless of revenue (in one example a company reached almost 400 employees on less than $1M in revenue before shutting down in 2022). [Elad Gil]

2022 proved a rude awakening. The pandemic boom had seen many companies crowned as unicorns, thousands of hires made, and new markets opened up. It also saw rampant inflation, and given that many governments coordinated their actions, inflation was not limited just to the USA.

In February 2022, Russia invaded Ukraine with devastating human and economic consequences. The attack placed a further squeeze on critical commodities. Costs that were already high, climbed even further, triggering inflation rates that far outpaced wage growth. In March 2022, in an effort to curb inflation, the Fed (and other countries equivalents) raised interest rates.

Interest rates determine how deeply you discount future cash flows, particularly for growth stage companies. When interest rates are higher, growth multiples compress and stock prices drop. April 2022 was a terrible month for stock markets and particularly bad for tech stocks - the Nasdaq dropped 13.3 percent the worst monthly performance since October 2008.

In response, venture capital funds and investors took to X (née Twitter) warning founders that tough times were coming and to buckle their belts. In May 2022, venture stalwart and occasional soothsayer Sequioa Capital published a 52-slide deck called “Adapting to Endure” laying out a litany of risks that would make it harder for founders to raise money and operate.

A cold-water realisation was washing over the startup community; with capital now constrained, the expensive race for growth no longer made sense. Where startups were once encouraged — even incentivised — to remain loss-making in order to gain market share. Now they were being told to focus on profitability. For many startups it was the beginning of the end, they’d built a balance sheet but no business and a reckoning was coming.

“It's not necessarily these founders' fault that they've been hearing one story from VCs about grow, grow, grow, and seemingly overnight, that story changed to the bottom line, bottom line, bottom line.” [Andrea Hippeau]

Companies moved to cut or freeze headcount and preserve cash at all costs. Growth rates slowed and later-stage startup valuations went into free fall. Perhaps the most visible example of this was Klarna, the European buy-now-pay-later provider. Exactly 12 months after being valued at $45.6B, its valuation dropped 85% to $6.7B.

2022 began with fears of a downturn after the historic highs of 2021, and ended with those fears fully realised. Median valuations trended down at every stage. Seed and Series A valuations ticked down only slightly, but later stages experienced much larger drops, reaching their lowest point since the 2010s. 

All in all, 2022 saw 1023 tech companies lay off 153,256 workers. Global startup investment declined by a third. The number of mega-rounds, in which startups raise $100m or more, fell by 71%. Unicorns, became rare again: the number of newborns contracted by 86%.

Liquidity all but disappeared, only $71.4B was generated in exit value, a 90.5% decline from 2021′s record of $753.2B. It was the first time exit value fell below $100B since 2016, with late-stage companies the hardest hit. Public offerings of VC-backed companies fell to a level not seen since the early 1990s, with just 14 public listings in the fourth quarter of 2022.

Hundreds of VC firms were now dead-men walking. Thanks to FOMO, they’d overspent and overpaid during the boom cycle – now the tide had gone out, they were left holding inflated paper returns with little chance of liquidity and no chance of persuading increasingly cautious LPs to invest in their next fund.

The era of easy money was well and truly over.

The gloom continued into 2023.

The scarcity of IPO exits, a pullback from nontraditional investors, ongoing economic headwinds and the collapse of Silicon Valley Bank (SVB) all contributed to the worst year for venture fundraising since 2015, with funds raising just $161B versus $307B a year ago. The slowdown was also evident from reports of reduced headcounts (including Anthemis, Sequoia Capital and Y Combinator) and missed fundraising targets (including Greycroft, Insight Partners and TCV).

Investors deployed capital more sparingly, with a higher bar at each stage. Global deal activity fell to ~$345B, down from $531B in 2022. Companies that saw high revenue multiples during the venture frenzy of 2020 and 2021 experienced hefty down rounds, particularly in once-celebrated sectors such as fintech, cryptocurrency, and the creator economy. Late-stage deals, in particular, became rarer, smaller, and cheaper.

In the tighter funding environment, many VCs favoured their existing portfolio over new deals, supporting their best-performing startups through extensions, bridges, or insider-priced funding rounds. In Q2 2023, bridge rounds comprised 38% of all fundings raised by companies whose latest priced round was a Series C.

Rebuilds take time as the industry slowly works through previous sins, & slowly regains confidence. Risk off is very fast. Risk-on is very slow. [Bill Gurley]

While most sectors declined, AI was one of the few winners. Global funding to AI startups reached close to $50 billion last year, up 9% from the $45.8 billion invested in 2022. Interestingly, most of the funding came from Microsoft, Nvidia and Google, who crowded out many traditional tech investors for the biggest deals in the industry.

Exits were tepid. The much vaunted $20B acquisition of Figma by Adobe was cancelled following pressure from European regulators. Large tech initial public offerings (IPOs) remained elusive. Public debuts from Arm, Klaviyo and Instacart sparked a glimmer of hope that public markets might be opening again. This was quickly extinguished by underwhelming after-market performance, reflecting ongoing mismatched valuation expectations between issuers and investors.

High-profile startups dropped like flies. WeWork collapsed into bankruptcy. The e-scooter company Bird, which became the fastest company to ever land a $1B valuation, is now worth less than $7m. SmileDirectClub, having raised $427m to revolutionise traditional orthodontics, ceased trading. Overall, more than 3,200 private venture-backed US startups went out of business.

The collapse of WeWork, a symbol of all the venture hubris of the last few decades seemed a particularly fitting end to this year (and this article).

Till next month.

A few weeks ago, I watched the founders of Humane unveil their new type of wearable, the AI pin, a screenless squircle that makes a big, bold sci-fi bet on being the device that will supersede the smartphone. 

My team were less than impressed … “I don’t get it?” … “why would you want this?” … “it’s literally Siri” … “so there’s no screen?” 

Their reaction reminded me of BlackBerry - a recent film that charts the extraordinary rise and spectacular fall of the eponymous smartphone manufacturer. In one dramatised scene, the Blackberry team are gathered around watching the iPhone unveiling at Apple’s 2007 keynote. Mike Lazaridis (the founder) looks at his BlackBerry and says, “Why would anyone want a phone without a keyboard?”

It didn’t end well for Blackberry. Apple and the iPhone took ever larger bites of their 45% hold on the cellphone market until it had virtually nothing left. 

This all got me thinking: why do we so often dismiss disruption?

Unsurprisingly, Clayton Christensen, author of “The Innovator’s Dilemma” and father of the term “disruptive technology” has a theory:

New technologies tend to undershoot what people actually want, and as a result are considered fun or niche by the general population.

Or, as Chris Dixon (Partner at a16z) puts it so succinctly in a brilliant article, “the next big thing will start out looking like a toy.” 

It doesn’t help that initial versions of disruptive technologies may seem unconventional or playfully experimental; they look and feel like novelties, so we treat them as such. Remember the early days of the internet? Cartoonish webpages, coloured scrollbars, autoplaying videos / music, hit counters and animated gifs …

When websites looked like this^ it’s hardly surprising that the economist Paul Krugman made his ill-fated Internet prediction:

“By 2005 or so, it will become clear that the Internet's impact on the economy has been no greater than the fax machine's.”

We also tend to forget many disruptive technologies undergo iterative development processes. The initial versions may lack the full functionality or features that make them viable for mainstream use. The first commercially successful portable computer, the Osborne 1, had no on-board battery and had to be plugged into the wall. The first commercially available handheld mobile, the Motorola DynaTAC, weighed over a kilogram and was affectionally known as The Brick.

It’s also the case that disruptive technologies often challenge existing norms and paradigms. They may introduce new ways of thinking or doing things that initially seem unconventional - like getting into a Cruise self-driving car or spending your leisure time in the Decentraland metaverse.

It’s why early technology adopters are typically geeks. They don’t buy products because of proven results, or an established track record. They buy because it’s new and innovative and because they want to tinker and try it out. Instead of being wary about no social proof, they are actually motivated by it.

You just have to look at the PC revolution. It was MITS, a small company that produced electronic kits for hobbyists, that made the first commercially successful personal computer - the Altair 8800 (1974). It wasn’t until much later (1983), when IBM launched their PC and Apple introduced Lisa that business computing eventually achieved widespread adoption (thanks largely to killer apps like word processors and spreadsheets.) Prices fell, performance improved and the real explosion began!

3D printing is a more recent example. What started as a perfect product for at-home DIY enthusiasts is now being used across industries, from Aerospace (eg. Relativity Space 3D prints rockets) to Construction (eg. Rebuild 3D prints buildings), and even the medical industry (eg. Xilloc 3D prints human bones!).

No discussion about geeks would be complete without mentioning gamers. In the process of writing this article, I came across The Gamer Disposition by John Seely Brown and Douglas Thomas in the Harvard Business Review. They share 5 key attributes that embody gamers - 3 of which are worth highlighting:

1. They thrive on change

   Gamers do not simply manage change; they create it, thrive on it, seek it out.

2. They see learning as fun.

   Gamers convert new knowledge into action and recognise that current successes are resources for solving future problems.

3. They marinate on the “edge.”

   Gamers desire to seek and explore the edges in order to discover some new insight or useful information.

With character traits like these, it’s unsurprising that gamers also tend to be early adopters of disruptive technologies. Whether it’s using cryptocurrency for in-game purchases, or NFTs to create, buy and trade characters, or joining virtual worlds (like Roblox) to connect, share and socialise.

Speaking of virtual worlds, another technology that has been adopted and popularised by gamers is Virtual Reality (VR). Thanks to increasing demand for immersive gaming (and the metaverse), headsets have gone from having to hang from a mechanical arm suspended from the ceiling to the Apple Vision Pro, a $3,499 pair of VR ski goggles.

So, it’s not that we necessarily dismiss disruption - it’s that the majority of us dismiss fringe behaviour and more often than not, lack the open-mindedness enjoyed by early adopters (the geeks and the gamers). So, the next time you look at something and think “it’ll never catch on”, reject your prejudice and remember your Virgil … “Beware Geeks Bearing Gimmicks”.

Till next month.

Last week, Nvidia (the global chip designer that’s fast becoming synonymous with Artificial Intelligence) invested in the pre-Series A round of generative AI startup Twelve Labs, their debut investment in a South Korean startup. 

It was also their 26th investment this year. Where other CVC / VC funds have pulled back on their deal numbers, Nvidia is plowing ahead and spending money like it’s going out of fashion. So what’s with the sudden flurry of deal making?

Well, as they say, make hay while the sun shines. Thanks to exorbitant demand for their AI-focused chips (such as the much coveted H100 GPU) in the second quarter of this year, Nvidia’s revenue reached 13.5Bn USD, doubling from the same period last year. For context, this is as much revenue as it used to haul in annually (as recently as 2020).

So, it’s only natural, with all that cash, that Nvidia should be spreading it around a bit, investing in their future (and making some AI friends along the way).

But here’s the rub… We don’t know whether Nvidia is investing cash in any of these deals - or if they’re investing stock, technology and services (or a combination of all of the above). It’s also worth noting that most of Nvidia’s portfolio companies are also their customers.

A cynic could argue that Nvidia is ‘round tripping’ - a nifty transaction whereby the company “sells an asset to another company, while at the same time agreeing to buy back the same or similar assets at about the same price”. In other words, the money Nvidia invests is used to buy GPUs off, you guessed it … Nvidia!

It gets even more ‘round trippy’ when you look at CoreWeave, another Nvidia portfolio company. CoreWeave specialises in cloud infrastructure for GPU-accelerated workloads. The recent boom in Generative AI has accelerated demand for their GPU Cloud to train and fine-tune models. As a result, the company has received a flood of cash - including (most recently) a $2.3Bn debt facility to “purchase compute to serve its customers”.

There’s nothing unusual about using debt to raise capital quickly, but what is unusual is that CloudWeave is using Nvidia’s much-sought after H100 AI GPUs as collateral. To summarise, Nvidia have invested in a startup, that is also a customer and a super user of their GPUs, and now these GPUs are being used as collateral for a debt facility that will be used to buy even more GPUs from (you guessed it) Nvidia … 🤔

In truth, this is probably less about financial shenanigans and more about king-making. CoreWeave’s privileged access to Nvidia chips gives them an edge on other cloud providers (Amazon, Google and Microsoft), all of whom are in the process of developing their own chips to try and end reliance on Nvidia. Even Google, which began creating competing A.I. chips more than a decade ago, relies on Nvidia’s GPUs for some of its work. By playing favourites with CoreWeave, Nvidia is creating a credible challenger to the cloud giants and fuelling future demand for its chips.

Whatever the case may be, Nvidia isn’t the first semiconductor company to use venture to stimulate the development of the ecosystem in which it operates. Intel Capital, the investment arm of the semiconductor giant, is the original gangster!

Back in 1991, long before corporate venture capital was in vogue, Intel realised it could benefit from enabling startups that made complementary products. Fuelling their demand could stimulate demand for Intel’s own microprocessor products. So, Intel invested in hundred’s of companies whose products would require increasingly powerful microprocessors, thereby driving sales for Intel’s chips.

They’ve since invested >$12Bn in >1,500 companies in >55 countries. Nearly 700 portfolio companies have gone public or participated in a merger - its best known investments include VMware (VMW), Citrix Systems (CTXS), Cloudera (CLDR), DocuSign (DOCU) and MongoDB (MDB).

It’s fair to say the ‘growing market strategy’ worked for Intel, but will it work for Nvidia? Is their venture frenzy a short-lived effort to capture market share before they lose momentum or are they establishing themselves as the Wizard of Oz for the AI industry as a whole? Only time will tell!

Till next month …

Subscribe now

It’s funny how some of the greatest polluters are also some of the least visible.

Take Cloud Computing (or the ‘Cloud’) as an example. To the average Joe, it’s as mysterious as it is innocuous. A ‘black box’ service that brings to mind an ephemeral realm of light, fluffiness and clear blue skies. This metaphor couldn’t be further from the truth.

Behind every Cloud there is a silver lining data centre - vast temples of computational power and storage, comprising miles of cabling, fiber optics, computer servers and more. All consuming huge quantities of electricity, water, air, metals, minerals and rare earth elements.

The environmental cost is absolutely staggering. The carbon footprint of the Cloud surpasses that of the airline industry. A single data centre can use the energy equivalent of 50,000 homes (and there are 8000 data centres globally). Annually, data centres collectively devour a staggering 200 terawatt hours (TWh) of energy, more than some nation states. When you include all networked devices (ie. laptops, smartphones and tablets) that make use of the Cloud, it accounts for >2% of global carbon emissions … it really should be called Carbon Computing!

Bitcoin is another silent environmental killer. To mint a bitcoin, you have to first “mine” it. Your computer is tasked with completing complicated equations that, if successfully done, validates and records transactions on the blockchain and rewards you with a newly created bitcoin and transaction fees. This is performed at an industrial scale and utilises high-performance computer chips called GPUs (graphics processing units) - and these GPUs require 10-15 times the energy used by a traditional CPUs (central processing units). As a result, Bitcoin mining uses more electricity than Norway and Ukraine combined.

So, what has this got to do with Artificial Intelligence? Well, AI runs in the Cloud and also uses GPUs (and lots of them); one report estimates that OpenAI's ChatGPT will eventually need over 30,000 GPUs. All of that computing is not just energy hungry; it’s also extremely thirsty.

AI consumes water in two ways – directly and indirectly – and it guzzles the most during the ‘training’ phase: the period in which AI is fed data and programmed how to respond.

Indirect consumption relates to the offsite generation of power used by the data centres. An example of this could be the water used in the cooling towers of coal-fired power stations. Direct consumption relates to the water used on-site by data centres for cooling purposes. Almost all of the power consumed by data centre servers is converted in heat - and this is exacerbated by the sheer size of the units. Heat must be relentlessly abated to keep everything whirring, 24 hours a day, every day. Water is used to cool the equipment and stop it overheating.

According to one estimate, ChatGPT gulps a 500ml bottle of water every time you ask a series of between 5-50 prompts or questions (the range depends on the location of the servers and the centre). The estimate also includes indirect water usage - such as cooling power plants that supply the data centres with electricity.

In its latest environmental report, Microsoft disclosed that its global water consumption spiked 34% from 2021 to 2022 (to nearly 1.7 billion gallons, or more than 2,500 Olympic-sized swimming pools), a sharp increase compared to previous years that outside researchers tie to its AI research.

Solving this problem is not easy, and some of the solutions are as wild as they are innovative - in 2018, Microsoft sunk a data centre off the Scottish coast and proved that underwater data centres are feasible, as well as logistically, environmentally and economically practical. Then there’s the ASCEND (Advanced Space Cloud for European Net zero emissions and Data sovereignty) programme. Their goal is to deploy data centres in Earth’s orbit, to demonstrate that data centres in outer space can substantially reduce their carbon footprint, by utilising solar energy beyond Earth’s atmosphere for power and the freezing vacuum of space for cooling.

Another option is to replace water-intensive and costly cooling processes with non-conductive fluids. These can be applied either via immersion cooling, where servers are submerged for immediate heat removal (see Submer, who are dunking servers in eco-goo to save the planet), or via spray cooling, where the Processer (the hottest component) is directly misted.

Ironically, AI might also be its own cure, whether its manufacturers using AI to design faster, smaller (and more energy efficient) chips, or data centre architects using AI to design and build lean and smart data centres, or data centre operators using AI and robotics solutions to improve energy efficiency and reduce carbon emissions. A Gartner report states that by 2025, half of cloud data centres will deploy advanced robots with AI and ML capabilities, resulting in 30% higher operating efficiency.

These solutions are much needed. Globally, about 4 billion people, or half the world’s population, are exposed to extremely high water stress at least one month a year. There is the potential implication that data centres are consuming water that should be reserved for at-risk communities. In the summer of 2022, Thames Water raised concerns about data centres around London using drinking water for cooling during a drought. This problem is only likely to get worse as the climate crisis deepens and weather patterns become more unpredictable.

As they say, the first step to recovery is acceptance - so it’s great to see the technology majors (Google, Microsoft, OpenAI etc.) have all acknowledged that this growing demand for AI carries a hefty environmental cost. Let’s hope they can apply their considerable creativity, ingenuity and resources to solve the problem!

Till next month …

Subscribe now

If I were to say “imaginary friends”, the first thing you’d probably think of is kids playing games and indulging in the land of make-believe - not a nascent multi-billion-dollar industry.

Thanks to recent advances in AI systems (see GPT-4), your computer is now able to generate text chats and voice conversations in an eerily lifelike fashion - contextually aware, grammatically perfect and totally devoid of emotion (insert lawyer joke here).

The recent proliferation of large language models has generated a slew of conversational use cases as varied as they are terrifying  awesome. They range from Air.ai replacing customer service reps with an AI that can hold 10-40 minute phone-calls, to the Khan Academy creating AI tutors for students, to Seoul-based Deep Brain AI letting you converse with your deceased loved ones.

If this all sounds suspiciously like Black Mirror, you’d be right. But Charlie Brooker isn't the only accurate film futurist - 10 years after Joaquin Phoenix fell in love with his AI companion, played by Scarlett Johansson in the Spike Jonze film “Her” - conversational AI is starting to make inroads into our emotional lives. 

One example is CarynAI, a voice chatbot created by influencer Caryn Marjorie. Users (unsurprisingly 99% men) pay $1 per minute to talk to an AI version of Caryn— a “virtual girlfriend”. According to a tweet by Justine Moore, a partner at venture capital firm A16z, Caryn made $72,000 in the first week of launch.

And this is not fringe behaviour. Replika (another digital companion app), has had over 10 million downloads, 2 million users and >250,000 subscribers paying for “an AI companion who cares”. Amusingly, back in Feb, Replika switched off their ERP functionality (erotic role playing, not enterprise resource planning), but have just relented to pressure from their audience and switched sexting back on.

At best, it's a peccadillo, at worst, it's weaponised. As intelligence analyst Remaya M. Campbell opinions; as AI chatbots blur the line between intimacy and secrecy, users could unwittingly divulge sensitive information - creating a new class of insider threats. 

Chatbot honeypots aside, it’s clear that AI companions are here to stay, and are going to play a very real, and very prevalent role in our personal lives. According to Mustafa Suleyman, the co-founder of Google DeepMind and Pi, a personal AI:

“Over the next few years, millions of people are going to have their own personal AI [and] in a decade everyone on the planet will have a personal AI.” 

Welcome to the age of robot love, where the heart wants what the algorithm decides.

Written by Alex Sainty