The Irish Elk, the largest deer to ever grace the Earth, is neither Irish nor an elk.
You can blame Scottish naturalist Thomas Molyneux for the misnomer. Upon finding a set of massive antlers and bones in an Irish peat bog, Molyneux mistook the bones for an elk because the antlers were so large.
The Irish Elk antlers, spanning up to 12 feet and weighing nearly 100 pounds, are something to behold. They became prized collector's items, displayed by Charles II and even Napoleon. But more importantly, the antlers played a pivotal role in our understanding of extinction.
At a time when extinction was not yet obvious, the Irish Elk's fossils, with their unmistakable antlers, helped George Cuvier, the founding father of paleontology, to prove extinction was real.
Still, the most fascinating aspect of the Irish Elk's story is the theory behind its extinction: those majestic antlers.
Integral to sexual selection, the antlers’ immense size and weight may have hindered the Irish Elk's mobility and imposed a heavy burden on their bodily resources. Amid changing habitats and increased competition, these once advantageous antlers ultimately sowed the seeds of the species' demise.
What does the story of the Irish Elk tell us?
For me, the Irish Elk is a cautionary tale about misallocation of resources. An overinvestment so bad that it may expose or otherwise create vulnerabilities that could jeopardize our long-term survival.
Over a long enough time horizon, the misallocation of resources toward something seemingly important, even beautiful or thrilling, can counterintuitively be the thing that leads to our destruction.
atoms and bits
Combatting the misallocation of resources has become a bit of a rallying cry for the United States government over the past few years.
Most obviously was how the pandemic revealed supply chain fragility caused by an overreliance on globalization that optimized for efficiency, rather than resilience.
Less obvious is the way that our public policy choices have led to another major misallocation of capital and talent in innovation that is optimized for investing in digital products and services, rather than breakthroughs in manufacturing and industrialization to support it.
The problem is that all of those digital products and services rely on the manufacturing and industrialization of semiconductor chips, which happens in Taiwan.
Taiwan, led by companies like the Taiwan Semiconductor Manufacturing Company (TSMC) and other foundries, is a global powerhouse in semiconductor manufacturing.
As advances in technology like AI expand, the demand for high-performance, energy-efficient chips skyrockets. As a result, the tech industry's dependence on Taiwan's semiconductors becomes even more pronounced at each stage of the technological acceleration that’s already happening.
The long term misallocation of resources in investing in atoms over bits has meant we have never been more vulnerable to rapid geopolitical and economic changes underway, particularly in Taiwan.
And the problem is it’s not possible for us to just wave a magic wand domestically–or more concretely for the President to invoke the Defense Production Act–and suddenly have America become a major semiconductor manufacturer overnight. That is a process that will take years, and may require more holistic changes to our public policies and regulations in the meantime, in order for that to come to fruition.
To that end, sadly it has been eight months since the CHIPS Act was signed into law, and yet we have basically nothing to show for it. What was viewed as our major piece of industrial policy strategy of the past five years has succumbed to red tape.
In the words of former Obama advisor Steve Rattnor, “Perhaps the biggest challenge confronting the chips team is that, under the National Environmental Policy Act, projects built with federal assistance are subject to lengthy reviews, which over the past decade have averaged 4½ years.”
Read that again. He’s talking about how long the review process is, the process for what is required before you start actually building. The building part also takes years.
fools gold
When I was in the Bay Area a few weeks ago, I got the chance to speak to one of the most dynamic CEOs in tech. He said he was excited right now, that there was so much opportunity to build.
And I asked him: do you mean like IRL or are you talking about AI and the digital world?
He laughed and said: “Digital. The physical world is just too frozen to do anything meaningful at scale.”
He mentioned he would maybe write more about that observation, and I hope he does.
But what I took away from the conversation with a person who is so hyper optimized in spotting massive growth opportunities is that if the physical world is too difficult to innovate and build, we are going to need to do much more than passing a CHIPS Act to prepare for the future that we are about to face.
The semiconductor industry will need to experience what the last decade was like for the United States in transforming our domestic energy production. It will need advances in technologies and support from the federal government to move faster for that type of manufacturing and production.
We need to reallocate capital and investment, namely through massive R&D tax credits, to support those who are building to accelerate the development of American semiconductor manufacturing, as well as finding new ways to support digital processing with fewer chips, less power, and more efficient computer systems to optimize for data movement.
There are startups working in this space, like Esperanto Technologies or Luminous Computing, just to name a few. But the point is, the federal government needs to work more closely with founders and investors to make sure that in allocating our talents and resources towards the historic wave of AI innovation, that we don’t forget the domestic semiconductor manufacturing and innovation that’s required to support it.
Because right now, it seems like AI looks like some beautiful, fascinating, but also ominous Irish Elk antlers.