The Tech Sales Newsletter #54: The AI conversation that never happened
Sales anon,
This month is a bit slow across the industry with many decision makers being away on vacation. After a soft first half, it seems like everybody is taking a big breath before the bonanza of activity that will play out over September, October and November.
Interestingly enough, that leaves some time for tech insiders to go on podcasts and private events. In this case we got a bit of a surprise, with a very candid and open conversation being published. It originally happened back from April with Eric Schmidt as a guest speaker. The talk was part of Stanford's ECON295 course led by the industry and academia insider Erik Brynjolfsson:
Source: ECON 295 course syllabus
ECON 295: The AI Awakening: Implications for the Economy and Society
This course will explore how the advances in AI can and will transform our economy and society in the coming years. Each week, we will learn from a guest speaker at the frontier of AI, economics, government or industry, read the relevant research, and discuss the implications. Primarily for graduate students in economics, business or computer science.
This is quite the roster, to say the least. Now Eric's lecture was in a conversation form and got quite a lot of attention because of two statements, one around Google's lack of desire to win and another one around copying code to launch new products.
I would say that these were not even the most interesting pieces, but rather a part of a bigger insider view at the intersection of tech, investment, AI, politics and the next wave of computing. The lecture was quickly taken down and scrubbed off the internet, so I'll share below the key insights from a tech sales perspective.
The insider view of AI
The conversation starts with Eric asking the audience several technical questions:
What is a million token context window?
What is text to action?
Why is NVIDIA worth $2 trillion and nobody can catch up?
These points frame the full discussion going forward:
A million token context window means that you can essentially prompt LLMs with the code of enterprise grade applications or complex science. Algorithms will be able able to deduct new insights and actions trough this extremely large context, including self-learning.
Text to action means something we already do, which is tell the thing to do the thing. Another way of thinking about it is text to python - natural language instructions resulting in code.
NVIDIA is worth $2 trillion not only because of hardware but also because of software, namely CUDA. CUDA is the default language right now for machine learning at a scale and the vast majority of software is heavily optimised for it.
Now let's dig into the key insights:
That's where we are with respect to agents, there are people who are now building essentially LLM agents. And the way they do it is they read something like chemistry. They discover the principles of chemistry, and then they test it-- --and then they add that back into their understanding, right? That's extremely powerful. And then the third thing, as I mentioned, is text action. So I'll give you an example. The government is in the process of trying to ban TikTok. We'll see if that actually happens.
If TikTok is banned, here's what I propose each and every one of you do. Say to your LLM the following, make me a copy of TikTok. Steal all the users. Steal all the music. Put my preferences in it. Produce this program in the next 30 seconds. Release it. And in one hour, if it's not viral, do something different along the same lines. That's the command. Boom, boom, boom, right? You understand how powerful that is?
If you can go from arbitrary language to arbitrary digital command, which is essentially what Python in this scenario is, imagine that each and every human on the planet has their own programmer that actually does what they want as opposed to the programmers that work for me who don't do what I ask, right?
The reason why you need to access the original content yourself than relying on media reports is precisely because the sales alpha is in the full context, not snippets.
The part about copying the code of TikTok and launching competitors made the news because it demonstrated how Silicon Valley insiders really approach new markets and technologies - move fast and let the lawyers deal with the consequences if it's valuable enough.
The real alpha however is understanding the synergy between LLMs and enterprise ML allowing for deep learning (which also includes testing and experimenting, then applying it back in the code) and text-to-action. If you can ask the LLM to do things, even if it's not possible from the first try you can keep providing feedback until it is able to do it. So then we can move away from "coding assistant" to essentially a programmer in your pocket.
So imagine a non-arrogant programmer that actually does what you want, and you don't have to pay all that money to. And there's infinite supply of these programmers. And this is all within the next year or two? Very soon. Those three things, and I'm quite convinced it's the union of those three things, it will happen in the next wave.
Now remember, this talk happened at the end of April. Since then, in June we had the Claude 3.5 launch with a 200k token context window and significant jumps in it's ability to generate Python code which the end users can literally just copy paste and run advanced commands or applications (hint: I'm doing it and if I can do it, so can you).
So at the moment, the gap between the frontier models, of which there are now only three, I'll review who they are, and everybody else appears to me to be getting larger. Six months ago, I was convinced that the gap was getting smaller, so I invested lots of money in the little companies. Now I'm not so sure. And I'm talking to the big companies, and the big companies are telling me that they need 10 billion, 20 billion, 50 billion, 100 billion. Stargate is 100 billion, right? They're very, very hard.
I talked-- Sam Altman is a close friend. He believes that it's going to take about 300 billion, maybe more. I pointed out to him that I'd done the calculation on the amount of energy required-- --and I then, in the spirit of full disclosure, went to the White House on Friday and told them that we need to become best friends with Canada because Canada has really nice people, helped invent AI and lots of hydropower. Mm-hmm.
Because we as a country do not have enough power to do this, the alternative is to have the Arabs fund it. And I like the Arabs personally. I spent lots of time there, right? But they're not going to adhere to our national security rules. Whereas Canada and the US are part of a triumvirate where we all agree-- So these $100 billion, $300 billion data centers electricity starts becoming the scarce resource?
The conversation then turns towards the real activity behind the scenes, namely the idea that the insiders have deep understanding of the real impact the technology is having, so they are much more focused now around the logistics of making it happen.
So Eric is having a lot of conversations with the key players at the main models and he is also talking to multiple governments on how to solve both the funding and the electricity problem.
I'm no longer a Google employee. Yes. In the spirit of full disclosure. Google decided that work life balance and going home early and working from home was more important than winning.
And the startups, the reason startups work is because the people work like hell. And I'm sorry to be so blunt, but the fact of the matter is if you all leave the university and go found a company, you're not going to let people work from home and only come in one day a week if you want to compete against the other startups. When-- in the early days of Google, Microsoft was like that. Exactly. But now it seems to be-- There's a long history of, in my industry, our industry, I guess, of companies winning in a genuinely creative way and really dominating a space and not making the next transition. It's very well documented.
And I think that the truth is founders are special. The founders need to be in charge. The founders are difficult to work with. They push people hard. As much as we can dislike Elon's personal behavior, look at what he gets out of people. Mm-hmm. I had dinner with him, and he was-- I was in Montana. He was flying that night at 10:00 PM to have a meeting at midnight with X.AI, right? think about it. I was in Taiwan. Different country, different culture. And they said that-- there's this TSMC who I'm very impressed with, and they have a rule that the starting PhD's coming out of-- their physicists work in the factory on the basement floor. Now, can you imagine getting American physicists to do that? The PhDs. Highly unlikely. Different work ethic.
And the problem here, the reason I'm being so harsh about work is that these are systems which have network effects. So time matters a lot. And in most businesses, time doesn't matter that much, right? You have lots of time. Coke and Pepsi will still be around, and the fight between Coke and Pepsi will continue to go along. And it's all glacial, right? When I dealt with telco's, the typical telco deal would take 18 months to sign, right? There's no reason to take 18 months to do anything. Get it done.
It's just we're in a period of maximum growth, maximum gain. And also it takes crazy ideas. Like, when Microsoft did the OpenAI deal, I thought that was the stupidest idea I'd ever heard. Outsourcing essentially your AI leadership to OpenAI and Sam and his team. I mean, that's insane. Nobody would do that at Microsoft or anywhere else. And yet today, they're on their way to being the most valuable company.
This was the statement that got the most attention and got the videos pulled - basically stating that Google lost it's desire to win and take actual risks, which led to an environment where companies that are driven by more motivated and aggressive leaders are able to get a real edge in the market.
The key part here is, we are in a period of maximum growth, maximum gain. The reason why he believes this ties back to one of the core thesis of this newsletter - computing power is a limited resource and at the scale that a true market leader needs to operate, in order for them to become that leader they will essentially acquire significant proportion of that computing power.
So the laggards are not only going to be "behind", they'll be starved to death.
So as you know, I lead an informal ad hoc, non-legal group. That's different from illegal. Exactly. Just to be clear.
Which includes all the usual suspects. Yes. And the usual suspects over the last year came up with the basis of the reasoning that became the Biden administration's AI Act, which is the longest presidential directive in history. You're talking to the Special Competitive Studies Project? No, this is the actual act from the executive office.
OK. And they're busy implementing the details. So far, they've got it right. Mm-hmm. And so for example, one of the debates that we had for the last year has been, how do you detect danger in a system, which has learned it, but you don't know what to ask it? Mm-hmm.
OK. So in other words, it's a core-- it's a sort of core problem. It's learned something bad, but it can't tell you what it learned, and you don't know what to ask it. And there are so many threats, right? Like, it learned how to mix chemistry in some new way, but you don't know how to ask it. And so people are working hard on that. But we ultimately wrote in our memos to them that there was a threshold, which we arbitrarily named as to the 26th flops, which technically is a measure of computation that above that threshold, you had to report to the government that you were doing this. And that's part of the rule. The EU, to just make sure they were different, did it at 10 to the 25. Yeah. But it's all kind of close enough. I think all of these distinctions go away because the technology will now-- the technical term is called federated training, where basically you can take pieces, and union them together. Mm-hmm. So we may not be able to keep people safe from these new threats.
The flip side of long context windows (self-learning) and text-to-action (language to python) is that we don't actually fully understand what's going on inside of the models. So if tomorrow ChatGPT 5 is able to make chemistry discoveries if left to operate a lab over night, we will not really understand how is that altering the code behind it. Eric and other insiders are spending a lot of time trying to lobby governments into this place where the work on developing new models can continue, but there are safeguards in place.
This is a bit different than the whole discussion around AGI because it's very practical - if the model can be used to trigger nuclear fusion with different components than what the existing science understands, then somebody could access these components and make a nuclear bomb. Which leads us of course to the next reveal, which is that Eric is now an arms dealer:
So I worked for the Secretary of Defense for seven years and tried to change the way we run our military. I'm not a particularly big fan of the military, but it's very expensive. And I wanted to see if I could be helpful. And I think in my view, I largely failed. They gave me a medal. So they must give medals to failure or-[LAUGHTER] --you know, whatever.
But my self-criticism was nothing has really changed, and the system in America is not going to lead to real innovation. So watching the Russians use tanks to destroy apartment buildings with little old ladies and kids just drove me crazy. So I decided to work on a company with your friend Sebastian Thrun, as a former faculty member here, and a whole bunch of Stanford people. And the idea basically is to do two things, use AI in complicated, powerful ways for these essentially robotic war. And the second one is to lower the cost of the robots. Now you sit there and you go, why would a good liberal like me do that? And the answer is that the whole theory of armies is tanks, artillery and mortar, and we can eliminate all of them. And we can make the penalty for invading a country, at least by land, essentially be impossible. It should eliminate the kind of land battles.
Well, this is-- related to your question, is that does it give more of an advantage to defense versus offense? Can you even make that distinction? Because I've been doing this for the last year, I've learned a lot about war that I really did not want to know. And one of the things to know about war is that the offense always has the advantage because you can always overwhelm the defensive systems. And so you're better off as a strategy of national defense to have a very strong offense that you can use if you need to. And the systems that I and others are building will do that. Because of the way the system works, I am now a licensed arms dealer. So computer scientist, businessman, arms dealer.
There is a lot to unpack here, but it ties back to what the day-to-day of tech insiders can mean and it goes way beyond code and money. Eric flat out admits that he is funding AI research focused on drone warfare and that his mission together with others would be to develop this to the stage that it becomes the new nuclear detergent.
This is not a surprise for anybody who has been paying attention in the last years and has been selling software in the Federal/EU NATO-related space. As an actionable from a tech sales perspective, I'll phrase this carefully - if you want to contribute to your country's national security strategy, selling cloud infrastructure software to the public sector is a viable path.
The consensus of my group that meets on every week is that eventually the way you'll do this so-called adversarial AI, is that there will actually be companies that you will hire and pay money to, to break your AI system. Like red team? So it'll be the red-- instead of human red teams, which is what they do today, you'll have whole companies and a whole industry of AI systems whose jobs are to break the existing AI systems and find their vulnerabilities, especially the knowledge that they have that we can't figure out.
Cybersecurity itself will be fundamentally changed by the developments in LLMs and Enterprise grade machine learning. Niches like pentesting will become fully AI driven but they'll also start impacting also how vulnerability research is done.
And then you also have to assume that there are tests for efficacy. So there has to be a way of knowing that the things exceeded. So in the example that I gave of the TikTok competitor, and by the way, I was not arguing that you should illegally steal everybody's music. What you would do if you're a Silicon Valley entrepreneur, which hopefully all of you will be, is if it took off, then you'd hire a whole bunch of lawyers to go clean the mess up, right? But if nobody uses your product, it doesn't matter that you stole all the content. [LAUGHTER]
But you see my point? In other words, Silicon Valley will run these tests and clean up the mess. And that's typically how those things are done. So my own view is that you'll see more and more performative systems with even better tests and eventually adversarial tests, and that will keep it within a box. The technical term is called chain of thought reasoning. And people believe that in the next few years, you'll be able to generate 1,000 steps of chain of thought reasoning, right? Do this, do this. It's like building recipes, right? That the recipes, you can run the recipe, and you can actually test that it produced the correct outcome. And that's how the system will work.
If the machine learning algorithms will evolve into essentially blackboxes, hallucinations will remain a problem (if more limited). For the purpose of building complex systems from these models, one workflow would be building reasoning steps that guarantee specific outcomes (i.e. hallucination free). This is more complex than what RAG is, where we essentially tell the models to only retrieve data from a specific set of documents.
There's a company called Mistral in France. They've done a really good job. And I'm obviously an investor. They have produced their second version. Their third model is likely to be closed because it's so expensive, they need revenue, and they can't give their model away. So this open source versus closed source debate in our industry is huge. And my entire career was based on people being willing to share software in open source. Everything about me is open source. Much of Google's underpinnings were open source. Everything I've done technically. And yet it may be that the capital costs, which are so immense, fundamentally changes how software is built.
One of the least well understood trends in the tech industry now is how many open-source companies are having to start shifting their business models towards closed source. This is not just an "AI company" problem, but rather impacts key players across the stack. This is relevant when evaluating companies to work for, because if open-source is seen as key part of their identity and strategy, they might simply not be able to scale and deliver outcomes in any way that's competitive.
To finish off this article:
So that's an agent model. And I think the text to action can be understood by just having a lot of cheap programmers, right? And I don't think we understand what happens, and this is again, your area of expertise, what happens when everyone has their own programmer. And I'm not talking about turning on and off the lights. I imagine, another example, for some reason, you don't like Google, so you say, build me a Google competitor.
Yeah, you personally-- build me a Google competitor. Search the web. Build a UI. Make a good copy. Add generative AI in an interesting way. Do it in 30 seconds and see if it works, right? So a lot of people believe that the incumbents, including Google, are vulnerable to this kind of an attack.
If LLMs can copy and generate code, then they can create applications. If they can create applications simply because we asked them to, how many software business today are truly "safe"?
For me the answer comes back to cloud infrastructure software. Just having an idea and the code to make this idea happen will not be sufficient to survive.
The winners of tomorrow will also have the hardware to keep it running. The low end of complexity in software will die out, replaced by complex applications that can only operate on reserved hardware.
The real tech business moat.
