Inner Loop Capital's Ad Interim newsletter explores AI trends: 1) AI in Digital Transformation, 2) AI Infrastructure emergence, 3) Core compute shift. Justin Label envisions "Hyperscaling-as-a-Service" in Infrastructure 5.0.
Dear Inner Loop Community:
Without a doubt the leading topic among founders and VCs for the last several months (except for a few weeks when we all became experts on bank balance sheets) has been Artificial Intelligence. The subject is potentially so far-reaching and impactful that I have decided to devote a recurring portion of my community correspondence exclusively to AI. Today I am launching an occasional newsletter called Ad Interim, a Latin phrase meaning “for the meantime”. I believe that applications of AI to software, tech investing, start-ups, and venture capital are currently moving so quickly that any investment roadmap or insight must be deemed to be a placeholder, regardless of how well-informed it is. As a result, these dispatches, as is always the case in venture capital, reflect my best thinking on the day they are offered. They are subject to – and almost certainly will – change.
But let us begin with that with which we are certain:
If you are a close friend of the firm – if you know my work well – then your suspicions were aroused before you finished the section above. Photonic Computing, Robotics, and Medical Transcription? This doesn’t sound much like Inner Loop’s core focus on Enterprise Infrastructure technologies like Cybersecurity, Cloud Technologies, and Data Science. If this struck you as odd, then congratulations. You know Inner Loop Capital better than GPT-4 does.
Nearly the entire section in blue italics above is simply false. So much so, it bears almost no connection to reality. The blue section above is the confident hallucination of ChatGPT to the following prompt:
Please write three paragraphs following this opening sentence: “Inner Loop Capital has been investing in AI start-ups since the firm’s inception.” Please include as much detail as possible, including naming specific portfolio company investments.
I appreciate the attribution from ChatGPT that I identified these AI companies, got conviction in them, and prevailed upon the founders to accept our capital. They all sound very compelling – although I now suspect I perhaps should not trust what I am told about them! At least all three companies seem to exist, although the description of Alonics bears no relation to any company I can find using that name.
The people who receive my letters will fall along a wide spectrum of experience with AI. Those of you with significant experience will know that it is no great trick to have generated the hallucinations above. (Even for GPT-4. And yes, I paid OpenAI $20/mo just to give you the above paragraphs.) I do not mean to imply that the publicly available AI technologies have no value, and certainly not that the imminent advances won’t be disruptive. Quite the opposite, which is why I am launching an entire series of letters devoted to this topic.
For the casual observer, however, I think it is wise to keep your skepticism intact in the face of some currently breathless headlines. Today’s available technology is not currently an all-knowing researcher, ready to replace our hard-earned judgment with instantly accurate analysis. Today’s systems strike me as always confident, and often wrong. Some professions reward that, so maybe those are most at risk from today’s tech. But for now it is important that we do not trust everything it generates.
Compelling for Simpler Tasks
There are some tasks I am comfortable giving to AI today. Here is one, from which you can easily extrapolate numerous valuable use cases.
Prompt: Please type this up into a formal email with proper business etiquette: “Can’t make meeting, Sarah will resched, Tues 18 or Thurs 20 best, afternoon my time both. Zosia moving project ahead meanwhile. Sorry & thx.
ChatGPT (GPT-4) response:
Subject: Meeting Rescheduling Request and Project Update
Dear [Recipient’s Name],
I hope this email finds you well. Unfortunately, I am unable to attend the previously scheduled meeting. In light of this, my colleague, Sarah, will be reaching out to you to reschedule. Our preferred dates and times for the meeting are as follows:
- Tuesday, April 18th, in the afternoon (my local time)
- Thursday, April 20th, in the afternoon (my local time)
We hope that one of these options works for your schedule. Please let us know if you have any preferences or alternative suggestions.
In the meantime, I wanted to provide a brief update on the progress of the project. Zosia is actively moving the project forward and will ensure that everything remains on track during this period.
I apologize for any inconvenience caused by my unavailability and truly appreciate your understanding. Thank you for your flexibility, and I look forward to our rescheduled meeting.
Best regards,[Your Name]
[Your Email Address]
Prepare for What is Likely, Keep an Eye on Tail Risks
I’m very conscious that my comfort level with AI applications may be backward-looking by several months. And that in a time of rapid change, that may make me a luddite. I do not doubt that the power of these technologies comes with substantial tail risk – on both the positive and negative sides of the axis. While not certain, I now believe that the following is sufficiently likely that I am acting as if it is true:
AI tech that is generally available in the near term is going to change everything about the sectors and businesses I care about and invest in. As a steward of LP’s capital, and as an advisor to the Inner Loop portfolio founders, I have a duty to get as smart as possible on the directions, capabilities, and risks of these technologies as quickly as possible. It is an unacceptable risk for any tech investor or entrepreneur to be a late adopter (or engager) of these technologies. They are potentially too important and moving too quickly to only stand on the sidelines and let the dust settle.
I do not presume that the current generation of natural language interface Large Language Models like ChatGPT will be the dominant, or perhaps even an important, part of that future. But I do believe that the underlying technology, thoughtfully applied to hundreds of domains, adapted as needed for different workflows, and continuing to evolve at today’s seemingly breakneck pace will affect almost everything in the technology world.
Therefore, you will be hearing much more from me on these topics at a regular cadence. As the title of this series implies, all conclusions and investment theses are always Ad Interim. To preview Issue #2, I will be talking about why my approach to the hype wave of AI is so very different to my approach to the prior hype wave of crypto and blockchain. I may be right both times, once, or neither time. But I will be transparent about my approach.
One major reason that I will be devoting the bulk of my research and writing to AI for the foreseeable future has to do with the role I believe I serve to a significant share of my LP base. Looking at the investors that have trusted Inner Loop with their capital commitments, I am often one of just a few connections for them to the start-up and venture capital industries. I feel this gives me a heightened duty to form coherent views of this important technology trend, share them with my community, and invest in a way that comports with them.
Even if this were not true – if I were just the “Enterprise Infrastructure puzzle piece” to my LP’s robust VC portfolios – I do not think I would have an option to just “stay in my lane” and let the AI VCs scuffle for the winners of the next AI wave. In all aspects of tech investing, I believe AI is going to drive new competitive advantages and risks. And I believe this is especially true for Inner Loop’s core Enterprise Infrastructure focus areas: Cybersecurity, Cloud Tech, and Data Science.
Inner Loop is Now Operating on the Presumption that this is Already a Platform Shift
“Platform Shifts” are exciting times for start-ups and venture capital investing. New applications are unlocked and incumbents are unseated. New buying and usage habits are formed and new dominant companies with increasing economies of scale are created.
Mainframes to servers. Terminals to PCs. Home computing. Internet connectivity. Dial-up to Broadband. WANs to vWANs to SASE. Mobile. Walled Gardens to User Generated Content to Social Networking. On-Prem to ASPs to SaaS. Public cloud. Anti-virus to anti-malware to EDR.
So despite the fact that I began Ad Interim #1 with a hype-check reflecting some inherent skepticism, I think it is now a fair bet that AI will be the next platform shift in technology. My favorite Bay Area bumper sticker from circa 2002 was “Lord, Please Give Me Just One More Bubble”. Well, that wish was granted several times over and has gotten us where we are today. While I am not above hoping that the multiples on the assets I already hold go up, a more apt wish for long-term technology investors with available capital might be: “Please Give Me One More Platform Shift.”
To borrow from the movie “Margin Call”:
Inner Loop won’t be investing in new chip or fab technologies to unseat Nvidia. It’s highly unlikely we will be involved in any new foundational AI models. However, it is extremely likely that over the years we will be involved in whole companies who work to apply those models to the spaces we invest in. And it is nearly certain that every company we touch going forward will utilize or be affected by these technologies. Ad Interim Issue #1 contains my first open comments on these trends which I expect will consume our research, investigation, and investing for many years to come.
I am pleased to count you in Inner Loop’s extended network while we pursue this work. And I hope you feel I am well positioned to play a role in this new phase of the tech industry. After all:[This is the abridged (!) ChatGPT (GPT-4) response to the prompt: “Please compose a glowing bio of Justin Label.”]
Managing Director (truly)
Inner Loop Capital