There is no self-made tech giant and Europe should remember that
Another week, another EU-level initiative to claw back some control of the technology that runs our lives.
The European Commission this week unveiled the European Technological Sovereignty Package, a set of measures it says aims to strengthen the continent’s capacity in semiconductors, artificial intelligence (AI), cloud and open source software.
The package has four main components. The Chips Act 2.0 updates the 2023 original to target AI chips specifically, speed up permitting, and bring chipmakers closer to demand sectors like data centres and AI factories.
The Cloud and AI Development Act aims to triple Europe’s data centre capacity within five to seven years and create a single EU-wide framework for assessing cloud and AI sovereignty.
Similarly, the EU Open Source Strategy will seek to build on Europe’s existing contributor base by scaling open source alternatives in areas like cloud, cyber security, and AI.
Finally, there is a, frankly puzzling, Strategic Roadmap for Digitalisation and AI in Energy is intended to address how to sustainably integrate surging data centre energy demand into the grid (a fine goal) while using AI to make Europe’s energy system smarter and cheaper (which, frankly, sounds like boilerplate ‘shove-AI-into-it’ waffle to me, but we’ll see).
This matters because, having lazily ceded all advantage in computing and information technology to the US, Europeans have now had a rude awakening, faced as we are not only with an increasingly erratic US but one that is also aligning itself with increasingly hostile tech giants. Likewise, the ongoing rise of China (whatever structural flaws are being papered over) demonstrates not only the rise of another economic competitor, but also the proof that competition is, in fact, possible.
It may even be a virtual categorical imperative.
Consider some recent events: when Broadcom acquired VMware, it restructured the licensing arrangements that hundreds of European cloud providers depended on to serve their customers. When the deadline passed, most lost the ability to sell VMware-based services at all. Costs had risen by up to 1,500% since the acquisition, and the affected companies were reportedly too afraid of retaliation to speak out publicly.
It’s not just about the platforms either, be they desktop operating systems or the cloud. The problem gets right down to the level of ones and zeros.
Sovereign silicon
For instance, despite the stated desire to build a sovereign cloud, the fact that data centres run on Intel and AMD chips means that Intel’s Management Engine and AMD’s Platform Security Processor are a risk. After all, they are black box, binary blob machines piggybacking on CPUs, and fully outside the control of EU governments.
Can you really have a sovereign cloud without sovereign silicon? And what does sovereign silicon look like? As I noted previously, code, whether saved to SSD or etched in silicon, is notoriously opaque, even when you can read it.
As this column has long been long on complaints about Europe’s failures in the technology sector I won’t continue to rehash the arguments here.
However, a couple of points occur: firstly, the founding myth of Silicon Valley, the bootstrapping entrepreneur in a garage, is just that: a myth. Garage start-ups did exist, but could only do so because the wider US technology sector already existed and was strategically, and hence financially, dependent on US government contracts, primarily defence. In other words, state subsidies.
Secondly, from the 1970s onward, external capital was widely available. The first venture capitalists, the American Research and Development Corporation and Draper, Gaither, and Anderson, both predate this significantly, but in 1972 venture capital (VC) came to Silicon Valley in a big way in the form of Kleiner Perkins opening on Sand Hill Road, and it transformed the sector.
This obviously got to absurd levels with so-called ‘blitz-scaling’, growth at the expense of profit in the hope of cornering the entire market, during the post-2008 zero interest rate era. The combination of ‘free’ money and investors desperate for returns had obvious consequences. Nevertheless, the availability of capital does matter.
Against the US tech sector model, typical European companies of the 1980s, when all of today’s tech was being put in place, tended to work more like gigantic small businesses, expanding through organic growth or bank loans and other lines of credit. Little wonder, then, that most of the names are now tombstones.
In light of all of this, the Commission’s announcement is a welcome one, and complaints about state aid should simply be shrugged off as what they are: dishonest.
More private money will be needed, though.
Risk
A 1979 US reform enabled pension funds to invest in risky assets including VC. In the following eight years, the share of US pension funds in VC capital increased from 15% to over 50%, creating the risk-taking culture that the US economy is now renowned for. EU pension funds today invest just 0.018% of their total assets in venture capital, whereas US pension funds allocate 1.9%.
An increase would obviously help, then, but proportionally Europe would still lag behind the US given that US retirement income is much more likely to ride on the market (which is not in itself necessarily a social good, even if it does boost the amount of capital available to businesses).
Either way, if Europe really wants to chart its own path, what would be most interesting to see would be a culture change, where the default options for both software and hardware were not the standard American tech giants.
Getting off creaking and increasingly ropey software is easy enough at the end-user level, and that would be a good first step. A second might be to look at the server-bound enterprise software that is rather harder to shift away from, whether because it is on-premise and customised or intentionally designed for data-driven vendor lock-in.
More than that, though, we need a new spirit of research, in both the public and private sectors.
New and different platforms, ones built from a different perspective and on different underlying assumptions, whether about the pricing model, the languages and compilers, the nuts and bolts of filesystems, or even the registers, would afford the opportunity to not only match the US but to do something fundamentally and entirely different. We could even design technology fit for humans. Imagine that.
Subscribers 0
Fans 0
Followers 0
Followers