08. Altman Era Index

Note : The Altman Era Index (AEI) is a working hypothesis, a thinking tool, not a formal framework. It isn’t affiliated with, endorsed by, or representative of any organisation, institution, or frontier lab. It is simply my attempt to make sense of how startups survive in a world shaped by fast-moving model cycles

Who can benefit most : Founders / Investors/ Enthusiasts gauging the health of an AI startup and its claims to succeed.

For those who know me, you'd notice my obsession for Sam Altman. My obsession stems from curiosity to understand how he thinks & operates; It's important, because there are few who can predict the future and wield power to create it. I scraped all his blog posts [available at blog[dot]samaltman[dot]com written from 2013] and been reading during flight hours from May 25. I have 40 pages left [175 pages in total] but this idea jumped midway.

The Why : Long story short, why did I read all of Mr.Altman's blogs ?

The Flash point idea : I sensed a strong shift in his writing style after 2022. What Sam wrote in 2014 is shaping reality today, my thesis : His latest ideas can probably hint direction towards future. Example, his predictions about Reditt, shared locations, Bitcoin, Chat Interface has become the fabric of reality today. Altman describes how acceleration feels from the inside as an operator, investor, and participant.

With this belief, I asked GPT to analyse his blogs and create a word cloud. While the word cloud looks simple, the vocabulary noticeably shifts from growth, startup, wealth to alignment, energy, governance. The nouns matured. Somewhere between 2013 and 2023 the way he wrote about things started to mature into futuristic concepts. If you plot those words over time, you’ll notice three clear gradients:

The question I’ve been obsessed with since reading him: If Altman can describe the slope of acceleration before it happens or use his ideas to predict the direction of Open AI, how do we compare and future proof our own ideas using that as a compass? If I am building an AI Startup, what era of Altman am I building into and what are the biggest threats to my business?

That’s where the idea of the Altman Era Index (AEI) came from, a simple system to measure where a startup’s thinking belongs. Are you building for the next decade or the previous one?

Every AI company, no matter how novel, is held together by six invisible systems:

Energy calculations are difficult to measure for some, I've eliminated it out of the scoring system for ease. We are calling the above systems as "Nodes" in this article further.

To see what “balanced” looks like, I built a reference from the current frontier labs: OpenAI, Anthropic, ElevenLabs, and Gemini/DeepMind. Their combined profile forms a blue sphere of near-perfect equilibrium. Initially I had thought of only Open AI, but added other frontier labs to even out the modality score.

Node Average Score (0–10) is the physics of the frontier, the cost floor, the learning cadence, the energy ceiling.

B. Measuring yourself against it :

Founders can self-diagnose by answering some questions in all honesty: What’s your cost per 1K tokens? How long does feedback take to reach production? How often do you audit for bias or drift etc?

Initially I wanted founders to score themselves on a scale of 1 to 10 to each question but that might introduce some unknown bias into the scoring pattern, so I've minimised the answer options and assigned a broader range to each option to make it easier to answer and bias free.

The algorithm computes three things:

Startups don’t usually die of competition; they drown in asymmetry. When there is too much compute, not enough feedback. Too much scale, no alignment. Too much energy burn, zero sustainability. For Example: If your Alignment score falls 20 % below frontier, you’ll lose enterprise trust. If your Compute × Energy balance breaks, you’ll drown in cloud bills.

The algorithm also shows a resulting table that captures the risk % of each node in a table format. The frontier gap also shows how far you are from any frontier lab.

AEI produces a dashboard of your survival probability. It is calculated based on three core pillars.

Frontier Fit (FF) : How close your architecture is to the frontier labs. If you are too far from the frontier sphere in CE, IQ, FB, AL, GV then your base resilience drops.FF is assigned a 40% weightage in the Survival score.

Symmetry Index (SI) : How balanced your 5-node design is. A lopsided company (one strong node, one very weak node) collapses faster under pressure. This is more of an interdependent score between the five nodes you have. I've assigned 30% weightage for SI in the survival score.

Competitive Pressure Index (CPI) : How much the next Altman-era will punish your weak spots. A high CPI means the upcoming Altman era exposes your vulnerabilities harder with a 30% weightage in the survival score (inverted as 1 − CPI)

You can read the survival map like a cockpit:

Altman-Fit > 0.8 → healthy orbit.

Symmetry < 0.7 → fragile.

CPI > 0.15 → frontier labs can overrun your moat within 12 months.

A simple gauge rolls it all up:

Green > 80 = Resilient. Yellow 60–80 = Fragile. Red < 60 = Drowning.

Survival Score (0–100) = 0.4 × Fit + 0.3 × Symmetry + 0.3 × (1 − CPI)

Wrapper Economics in AEI captures three forces to reckon :

1. NET (Normalized Economic Throughput)

How big you are relative to the frontier labs. Small NET = labs barely notice you. Large NET means your business is exposed whenever they change pricing, quotas, or policies.

2. CER (Customer Exposure Risk)

How much a price cut or model upgrade from labs can compress your margins. If the lab halves its API cost, and you pass that to customers, you’re safe. If it destroys your pricing power the CER spikes.

3. FSR (Frontier Shock Risk)

This is the “earthquake” metric. It measures how destabilising new frontier model releases can be for your business or the age old fear : Will Open AI kill my startup overnight.

Wrapper Economics is, in essence is:

Every new frontier model release creates a shock wave that moves through the ecosystem and AEI models this using two simple ideas:

Frontier Velocity : How frequently the labs drop major upgrades per year.

Each model launch or shock cycle lowers your adjusted survival unless:

The Frontier Cycles Map is a projection of your startup’s survival score over the next 6 frontier model cycles. Think of it as a stress-test through time and not revenue.

The AEI isn’t a prophecy engine; it’s a mirror. It forces founders to think in systems, not slogans. To see their own asymmetries before the market does. To ask: If the frontier moves tomorrow, which part of me drowns first?

You can run the AEI from this Colab notebook here. I am not recording or saving your data anywhere. Be honest in your answers and let me know the results. Again please take it with a pinch of salt.

Altman once wrote, “Prediction is not foresight; it’s calibration.” The AEI is exactly that. A simple calibration tool. A way for every AI founder to ask some honest questions and reflect in theie alone time :

“Which AI decade am I really building for?"

"What are my weakest spots"

"How much is my business exposed to frontier shocks"