The Exponential Horizon: Why Short-Termism is Foolish Near the Singularity¶

Original Text¶

We are at one of the most profoundly asymmetric moments in history. The only move right now is to lengthen your time horizon and abandon short-termism entirely.

Being overly concerned about bubbles is foolish. Trying to time the market is foolish. Short term fluctuations and corrections will always happen but are entirely noise since we're in such close reach to the singularity. We are going to have runaway growth in AI/Robotics/Energy/innovation. We will have billions (or more) of AI agent workers, humanoid robots, data centers in space, multiplanetary colonization, vastly better medical therapies and fundamentally change the development speed and throughput of the number of technological breakthroughs across all domains within the next decade. We will compress more technological progress and economic growth in the next 2 decades than in the entire history of civilization combined.

We are already on the steep part of the J curve but it's hard to see when we zoom in to the day to day or week to week. 100% of Anthropic product code is now written by Claude. PMs have a team of virtual SWEs that almost transmute time. Companies that efficiently utilize AI have been increasing their product iteration speed not by single digit %, not double digit %, but triple digit %. And the capabilities of these tools are still improving at an ever faster rate. Whether we officially hit ASI in 2027 or 2029 doesn't really matter. It WILL happen. By the time it's officially declared, asset prices of the things you want to own will be many many times higher.

Quite feasibly, the next 3-10 years of real economic growth could register as 20-sigma events under any historical distribution. Growth that was previously considered near impossible, driven by second- and third-order changes that have no precedent. Traditional valuation models are unequipped to price these changes. The potential for runaway upside is so vast that it's very difficult to capture them with conventional present value calculations.

The velocity of wealth appreciation will be extremely startling just like it was when crypto first minted many billionaires and centimillionaires in a short period of time, but orders of magnitude more extreme. If not exposed, it will be very hard to buy into prices that are appreciating so quickly, but unlike previous bubbles, real economic value creation will be able to keep much better pace with vertical asset appreciation. Those that have operated with the understanding of the exponential horizon over the last 3 years have benefited greatly. If you have not adopted this understanding, it's not too late.

It's essential to always think about downside risk, but this is the largest upside risk the world has ever seen. Learn how to stomach risks for longer. This is not the time to trade. Investing generally outperforms trading for the vast majority of the population but the gap between the EV of trading vs investing will grow larger than ever. How much is the embedded call option of the singularity worth?

English Summary¶

Andrew Kang argues we're at the most asymmetric moment in history—approaching the singularity with runaway AI/robotics/energy growth that will compress more progress in 20 years than all of civilization combined—making short-term trading foolish and demanding radically extended time horizons, because the next 3-10 years could be 20-sigma events that traditional valuation models cannot price, and by the time ASI is officially declared, asset prices will already be "many many times higher."

The Core Thesis: We're Already on the Steep Part of the J-Curve¶

Evidence of acceleration (happening NOW): - 100% of Anthropic product code written by Claude - PMs have "virtual SWE teams that almost transmute time" - AI-efficient companies increasing product iteration speed by triple-digit % (not single, not double) - Tool capabilities improving at "ever faster rate"

Implication: We can't see the exponential curve because we're zoomed into day-to-day or week-to-week noise. But the inflection has already happened.

The 20-Sigma Event: Next 3-10 Years¶

What's coming (next decade): - Billions (or more) AI agent workers - Humanoid robots - Data centers in space - Multiplanetary colonization - Vastly better medical therapies - Fundamental change in development speed + throughput of technological breakthroughs across ALL domains

The compression:

"We will compress more technological progress and economic growth in the next 2 decades than in the entire history of civilization combined."

Statistical framing: - Next 3-10 years could register as 20-sigma events under any historical distribution - Growth previously considered "near impossible" - Driven by second- and third-order changes with no precedent

Why traditional models fail: - Traditional valuation models "unequipped to price these changes" - Runaway upside potential "very difficult to capture with conventional present value calculations"

ASI Timing is Irrelevant¶

The argument: - Whether ASI arrives in 2027 or 2029 "doesn't really matter" - It WILL happen - By the time it's officially declared, "asset prices of the things you want to own will be many many times higher"

The embedded call option: - The singularity itself is an embedded call option in all assets - Question: "How much is the embedded call option of the singularity worth?" - Answer (implied): More than any conventional valuation can capture

Wealth Velocity Will Be Extreme¶

Comparison to crypto: - Crypto "minted many billionaires and centimillionaires in a short period" - This will be "orders of magnitude more extreme"

Critical difference from past bubbles: - Unlike previous bubbles, real economic value creation will keep pace with vertical asset appreciation - Not just speculative mania—actual productivity explosions

The timing challenge: - "If not exposed, it will be very hard to buy into prices that are appreciating so quickly" - Implication: Get positioned NOW, or you'll be priced out

Strategic Implications: Abandon Short-Termism¶

What NOT to do: - ❌ Worry about bubbles ("foolish") - ❌ Try to time the market ("foolish") - ❌ Trade short-term fluctuations ("entirely noise")

What TO do: - ✅ Lengthen time horizon - ✅ Abandon short-termism entirely - ✅ Learn to stomach risks for longer - ✅ Invest, don't trade

Why investing >> trading (more than ever): - "Investing generally outperforms trading for the vast majority" - But the EV gap between investing vs trading "will grow larger than ever" - Because: Trading optimizes for noise; investing captures the exponential call option

It's Not Too Late¶

For those who haven't adopted this framing:

"If you have not adopted this understanding, it's not too late."

Evidence from the last 3 years: - "Those that have operated with the understanding of the exponential horizon over the last 3 years have benefited greatly" - Implication: The exponential acceleration has been underway since ~2023

The Asymmetry: Largest Upside Risk Ever¶

Risk framing: - "Essential to always think about downside risk" - BUT: "This is the largest upside risk the world has ever seen"

The asymmetry: - Downside: Traditional risks (recession, correction, bubble pop) - Upside: Singularity (20-sigma event, runaway growth, civilizational-scale compression) - Traditional risk management optimizes for the wrong tail

繁體中文總結¶

Andrew Kang 主張我們正處於歷史上最不對稱的時刻——接近奇點，AI/機器人/能源的失控性成長將在 20 年內壓縮超過整個文明史的進步總和——這使得短期交易變得愚蠢，並要求極端拉長時間視野，因為未來 3-10 年可能是傳統估值模型無法定價的 20-sigma 事件，而當 ASI 正式宣布時，資產價格已經「高出許多倍」。

核心論點：我們已經在 J 曲線的陡峭段¶

加速的證據（正在發生）： - Anthropic 100% 產品程式碼由 Claude 撰寫 - PM 擁有「幾乎能轉換時間的虛擬 SWE 團隊」 - AI 效率高的公司產品迭代速度提升三位數 %（非個位數、非雙位數） - 工具能力以「越來越快的速度」改善

含義： 我們看不到指數曲線，因為我們放大到每日或每週的噪音。但轉折點已經發生。

20-Sigma 事件：未來 3-10 年¶

即將到來（未來十年）： - 數十億（或更多）AI agent 工作者 - 人形機器人 - 太空數據中心 - 多行星殖民 - 大幅改善的醫療療法 - 所有領域的科技突破開發速度與吞吐量根本性改變

壓縮：

"我們將在未來 20 年壓縮比整個文明史總和更多的科技進步與經濟成長。"

統計框架： - 未來 3-10 年可能在任何歷史分佈下註冊為 20-sigma 事件 - 過去被認為「幾乎不可能」的成長 - 由沒有先例的二階、三階變化驅動

為何傳統模型失效： - 傳統估值模型「無法為這些變化定價」 - 失控的上行潛力「很難用傳統現值計算捕捉」

ASI 時間點不重要¶

論點： - ASI 在 2027 還是 2029 到達「並不重要」 - 它會發生 - 正式宣布時，「你想擁有的資產價格將已經高出許多倍」

嵌入式買權： - 奇點本身是所有資產中的嵌入式買權 - 問題：「奇點的嵌入式買權價值多少？」 - 答案（暗示）：超過任何傳統估值能捕捉的

財富速度將極端¶

與加密貨幣比較： - 加密貨幣「短期內造就許多億萬富翁和億萬富翁（centimillionaires）」 - 這將是「數量級更極端」

與過去泡沫的關鍵差異： - 與過去泡沫不同，真實經濟價值創造將跟上垂直資產升值 - 不只是投機狂熱——實際的生產力爆炸

時機挑戰： - 「如果沒有曝險，將很難買入快速升值的價格」 - 含義：現在定位，否則將被定價排除

策略含義：放棄短期主義¶

不要做什麼： - ❌ 擔心泡沫（「愚蠢」） - ❌ 試圖擇時（「愚蠢」） - ❌ 交易短期波動（「完全是噪音」）

要做什麼： - ✅ 拉長時間視野 - ✅ 完全放棄短期主義 - ✅ 學會更久地承受風險 - ✅ 投資，不要交易

為何投資 >> 交易（前所未有）： - 「投資對絕大多數人普遍優於交易」 - 但投資 vs 交易的 EV 差距「將比以往更大」 - 因為：交易優化噪音；投資捕捉指數買權

現在還不算晚¶

對於尚未採納此框架的人：

"如果你還沒有採納這種理解，現在還不算晚。"

過去 3 年的證據： - 「過去 3 年以指數視野理解運作的人已大幅受益」 - 含義：指數加速自 ~2023 年起就在進行

不對稱性：有史以來最大的上行風險¶

風險框架： - 「永遠思考下行風險是必要的」 - 但：「這是世界有史以來最大的上行風險」

不對稱性： - 下行：傳統風險（衰退、修正、泡沫破裂） - 上行：奇點（20-sigma 事件、失控成長、文明規模壓縮） - 傳統風險管理優化錯誤的尾部

Key Quotes¶

"We are at one of the most profoundly asymmetric moments in history."

"Being overly concerned about bubbles is foolish. Trying to time the market is foolish. Short term fluctuations and corrections will always happen but are entirely noise since we're in such close reach to the singularity."

"We will compress more technological progress and economic growth in the next 2 decades than in the entire history of civilization combined."

"100% of Anthropic product code is now written by Claude. PMs have a team of virtual SWEs that almost transmute time."

"Companies that efficiently utilize AI have been increasing their product iteration speed not by single digit %, not double digit %, but triple digit %."

"Whether we officially hit ASI in 2027 or 2029 doesn't really matter. It WILL happen. By the time it's officially declared, asset prices of the things you want to own will be many many times higher."

"Quite feasibly, the next 3-10 years of real economic growth could register as 20-sigma events under any historical distribution."

"Traditional valuation models are unequipped to price these changes. The potential for runaway upside is so vast that it's very difficult to capture them with conventional present value calculations."

"The velocity of wealth appreciation will be extremely startling... but orders of magnitude more extreme [than crypto]. Unlike previous bubbles, real economic value creation will be able to keep much better pace with vertical asset appreciation."

"If not exposed, it will be very hard to buy into prices that are appreciating so quickly."

"If you have not adopted this understanding, it's not too late."

"This is the largest upside risk the world has ever seen. Learn how to stomach risks for longer."

"This is not the time to trade... the gap between the EV of trading vs investing will grow larger than ever."

"How much is the embedded call option of the singularity worth?"

Personal Reflection¶

Why This Matters¶

This is one of the clearest articulations of exponential optimism I've seen—and it directly contradicts the recent macro analysis we've archived (plur_daddy's capital scarcity, Andy Constan's credit cycle risks). Three aspects make this significant:

1. Falsifiable near-term predictions

Most singularity theses are unfalsifiable ("it'll happen eventually"). Andrew makes specific claims: - Anthropic: 100% code written by Claude (verifiable NOW) - Product iteration speed: triple-digit % increases (measurable) - ASI arrival: 2027-2029 (2-year window)

If these don't pan out, the thesis breaks. This is intellectually honest futurism.

1. The 20-sigma framing is deliberate hyperbole (or is it?)

Calling next 3-10 years a "20-sigma event" is statistically absurd—under a normal distribution, a 20-sigma event has probability ~10^-88 (effectively impossible).

But that's the point: If the singularity happens, historical distributions are irrelevant. The framing forces you to think outside reference class forecasting.

The question isn't "is this literally 20-sigma?" but "are we in a regime where historical priors break?"

1. It's an investment thesis, not a prediction

Andrew isn't claiming certainty. He's making an asymmetry argument: - Downside: You hold assets, market corrects, you lose 20-50% - Upside: Singularity happens, assets 10x-100x in a decade

If you assign even 10% probability to the singularity scenario, the EV dominates. This is Kelly Criterion logic applied to civilizational-scale bets.

The Contradiction with Plur_Daddy / Andy Constan¶

We now have three contradictory theses in the archive:

Thesis 1 (plur_daddy): Capital Scarcity → Crypto Collapse - AI capex draining global capital - Deepest pockets (Saudis, Softbank) tapped out - Speculative assets (crypto) crushed as "tip of the spear" - Recommendation: Cash, wait for opportunities

Thesis 2 (Andy Constan): Credit Creation Regime → Asset Bifurcation - Private sector credit replacing central bank money creation - Large promises (AI capex, onshoring) must be funded via asset sales - Stock/bond divergence, speculative assets suffer - Recommendation: Selective equities, active risk management

Thesis 3 (Andrew Kang): Exponential Horizon → Buy Everything - Singularity approaching, 20-sigma upside imminent - Short-term noise irrelevant, timing is foolish - Real economic value creation will justify vertical asset appreciation - Recommendation: Extend time horizon, hold, don't trade

These cannot all be simultaneously correct. Or can they?

Reconciliation Attempt: Different Time Horizons¶

Possibility: All three are correct, just at different timescales:

Near-term (2026-2027): plur_daddy / Andy Constan are right - Capital scarcity real - Credit cycle stress real - Speculative assets (crypto) struggle - Volatility, corrections, asset sales

Medium-term (2027-2029): Transition period - AI capex starts paying off (productivity explosions) - ASI threshold crossed (2027-2029 per Andrew) - Valuation models break down - Asset prices start reflecting exponential future

Long-term (2030+): Andrew Kang is right - Runaway growth phase - 20-sigma economics - Historical distributions irrelevant - Vertical wealth appreciation

The investment implication: - 2026-2027: Survive the squeeze (cash reserves, manage risk) - 2027-2029: Position for exponential (accumulate assets) - 2030+: Hold and compound (don't trade the noise)

Falsification Criteria¶

Andrew's thesis breaks if:

1. Anthropic claim is false
2. If 100% code-by-Claude is marketing spin, not reality

3. Verifiable: Ask Anthropic employees, check GitHub activity

4. Triple-digit iteration speed is cherry-picked

5. If only a handful of companies see this, not broad-based

6. Need: Industry-wide data on product release velocity

7. ASI doesn't arrive by 2030

8. If 2027-2029 window passes without ASI declaration

9. (Though Andrew can claim "it arrived, just not officially declared")

10. Asset prices don't appreciate vertically

11. If next 3 years see flat/declining markets
12. But: Andrew would argue "zoom out, look at 10-year chart"

The unfalsifiable core: - "Real economic value creation will keep pace" is impossible to verify in advance - Past bubbles also claimed "this time is different" - The singularity is definitionally unpredictable

What I'd Watch¶

Leading indicators that Andrew is right:

1. Productivity data acceleration
2. GDP per capita growth rates (US, global)
3. Total factor productivity (TFP) growth

4. Labor productivity in AI-adopting sectors

5. AI deployment velocity

6. % of companies with AI-written code (Anthropic is data point 1)
7. Humanoid robot shipments (Tesla Optimus, Figure, etc.)

8. AI agent adoption (customer support, coding, research)

9. Energy/compute buildout

10. Data center construction rates
11. Nuclear/fusion investment flows

12. Space-based infrastructure announcements

13. Asset price behavior

14. Do corrections get bought aggressively? (Sign of "don't fight the exponential")
15. Do valuations expand despite fundamentals? (Market pricing in singularity premium)

16. Does crypto recover despite capital scarcity? (Exponential optimism >> liquidity)

17. Narrative shift

18. When does mainstream media adopt "singularity imminent" framing?
19. When do central banks acknowledge "traditional models broken"?
20. When does "it's not too late" become "you're too late"?

Leading indicators that plur_daddy / Andy are right:

1. Credit events
2. Oracle bond spreads continue widening
3. FDI promises fail to materialize (onshoring stalls)

4. Consumer credit contracts (no job creation from AI)

5. Asset liquidations

6. Sovereign wealth funds selling treasuries (funding AI capex)
7. Hedge fund redemptions accelerating

8. Crypto capitulation extends (no exponential recovery)

9. Policy response

10. Fed forced to ease despite inflation (credit crunch)
11. Or: Warsh tightens into weakness (misdiagnosis confirmed)

The Ultimate Question: Which Tail Are You Optimizing For?¶

Traditional risk management: Optimize for downside (recession, correction, bubble pop) - Result: Miss the 100x if singularity happens

Exponential risk management: Optimize for upside (singularity, 20-sigma event) - Result: Suffer 50% drawdown if capital scarcity persists

Andrew's framing:

"It's essential to always think about downside risk, but this is the largest upside risk the world has ever seen."

The asymmetry: - If he's wrong: You hold assets through a correction, recover eventually - If he's right: You capture civilizational-scale wealth creation

The embedded question:

"How much is the embedded call option of the singularity worth?"

This is asking: What % of your portfolio should be allocated to a 10-100x outcome with 10-30% probability?

Kelly Criterion says: Allocate proportional to edge. If singularity has 20% probability and 50x upside, you should be 100% allocated (because expected value dominates).

But Kelly assumes you can survive losing 100%. Most people can't. So the practical question is:

What's the minimum allocation that lets you capture exponential upside while surviving near-term capital scarcity?

Cross-Reference: Three Theses in Tension¶

Can all three be true? - Yes, if: Near-term pain (plur/Andy) → medium-term transition → long-term moonshot (Andrew) - No, if: Capital scarcity persists indefinitely (singularity delayed/canceled)

The key question: Is the singularity inevitable (Andrew) or contingent on surviving the capital squeeze (plur/Andy)?

Investment Implications (If You Believe Andrew)¶

What to do:

1. Extend time horizon to 10+ years
2. Accept near-term volatility as noise

3. Don't sell corrections ("foolish to time the market")

4. Maximize exposure to exponential upside

5. AI infrastructure (Nvidia, hyperscalers)
6. Robotics (Tesla, Figure, etc.)
7. Energy (nuclear, fusion)

8. Potentially crypto (if it survives capital scarcity and captures singularity premium)

9. Avoid trading

10. EV gap between investing vs trading "larger than ever"

11. Trading optimizes for noise; investing captures call option

12. Stomach larger drawdowns

13. "Learn how to stomach risks for longer"

14. 50% correction on path to 10x is acceptable (if thesis correct)

15. Ignore traditional valuation

16. "Traditional models unequipped to price these changes"
17. P/E ratios, DCF models assume historical distribution
18. If singularity happens, these are irrelevant

What NOT to do:

1. ❌ Hold cash waiting for correction (miss the exponential)
2. ❌ Trade volatility (optimize wrong dimension)
3. ❌ Apply historical risk management (wrong tail)
4. ❌ Wait for "official ASI declaration" (priced in by then)

The embedded assumption: - You can survive near-term volatility without forced selling - If capital scarcity forces liquidation, you miss the exponential

Risk: If plur_daddy / Andy are right and Andrew is wrong, you suffer 50-80% drawdown with no recovery.

Final Thought: The Unfalsifiable Core¶

Every singularity thesis has an unfalsifiable core:

If ASI arrives: "I told you so, you should have listened" If ASI doesn't arrive: "It's coming, just delayed, extend your horizon"

Andrew is more falsifiable than most (2027-2029 window, specific near-term claims). But the ultimate claim—"real economic value creation will keep pace with asset appreciation"—cannot be verified until after the fact.

The question isn't "is Andrew right?"

The question is: "What probability do I assign to this scenario, and how should I position?"

If 0%: Hold cash (plur_daddy) If 10%: Small allocation to asymmetric upside If 50%: Heavy allocation, accept volatility If 90%: 100% assets, lever up

Andrew's message: Even if you're uncertain, the asymmetry favors exposure.

"It's not too late."

But it might be soon.