Running Before Learning to Walk: The Data Behind the AI Industry's Two Deadlocks

By Dr. Tong Yin (殷彤博士) · Founder & CEO, InsightBridge Global LLC

Introduction

In 2026, on stages across Silicon Valley, executives from Meta, OpenAI, and Google take turns proclaiming that Artificial General Intelligence (AGI) is about to rewrite everything, that AI will replace human labor wholesale. Wall Street buys the story. The media amplifies it. Valuations soar.

Yet the moment you strip away the smoke and mirrors—generated by exorbitant “mercenary scientists” and endless venture capital—and confront the ground-level commercial reality, a starkly opposite conclusion emerges: the current AI industry is trapped in a dysfunctional loop of trying to run before it has learned to walk.

This is not a sentiment. It can be verified with two cold sets of facts:

• First, this development model cannot generate profits. Aside from a single exception, nearly all downstream AI giants are bleeding astronomical amounts of cash, while the four biggest cloud players just raised 2026 AI capex 77% year over year.
• Second, because they refuse to do the foundational grinding work, no model has reached true industrial-product grade. MIT’s latest research shows that 95% of enterprise AI pilots produced no measurable business impact.

Below, using data and facts, I unpack these two deadlocks one by one.

Part One: Unable to Profit — A Massive Loss-Making Game Sustained by Capital Transfusions

1. Mercenary Scientists: The Empty-Chair Game Behind $100 Million Contracts

To catch up with OpenAI, Meta has offered $100 million-level compensation packages to poach top Chinese scientists. Headline-shocking, but here is the real structure:

Table 1: Anatomy of a $100M Compensation Package

Source: standard Silicon Valley Base + Signing + RSU structure, uniform vesting over four years.

The industry’s real data:

• Median tenure of a top AI scientist at any one company: 1.5 to 2.5 years;
• OpenAI employees have a median tenure of only 16 to 18 months (Business Insider);
• Which means these headline contracts are never fully paid out.

A scientist walks away from Meta after 1.5 years with roughly $54 million, forfeits the remaining $40M+ in unvested stock—then joins the next giant, which covers the forfeited stock with an even bigger “buyout signing bonus.” They are the top mercenaries of the tech world, jumping every two years, forever cashing out mid-flight.

Wonderful for the individual scientist. Catastrophic for the industry:

• R&D goes fast-food: A trillion-parameter model takes only 6 to 12 months from concept to release. Scientists chase publications, benchmarks, architectural novelty—not shippable products.
• Discontinuity: The moment a lead scientist leaves, the successor rewrites the codebase. Billions of dollars of compute evaporate.
• Only takeoff, never landing: Once the architecture is up, the grinding work of edge cases, stability, and cost optimization—the exact work that turns a model into a product—is beneath them, and no one has time for it.

2. The Real P&L Ledger of Downstream Model Companies

Under this continuous burn of headline compensation and massive compute, here is what 2025–2026 actually looks like:

Table 2: OpenAI 2024 vs 2025 Audited Financials (Leaked)

Source: Ed Zitron / Financial Times / Fortune / Ars Technica, June 2026. Note: the $38.53B 2025 net loss includes $41.55B in one-time non-cash charges from the nonprofit-to-for-profit conversion; the comparable cash loss is closer to $8B. The $20.92B operating loss is the cleanest reflection of ongoing operations.

Table 3: Top AI Labs P&L Comparison, 2025–2026

Source: Fortune, Financial Times, CNBC, Mental Momentum Research, June–July 2026.

Table 4: The Hyperscaler AI Capex Arms Race

Source: Goldman Sachs, S&P Global, Yahoo Finance, Gate.com, June–July 2026. CreditSights estimates \~$450B of 2026 hyperscaler capex flows directly into AI infrastructure.

Goldman Sachs further projects combined hyperscaler capex of $5.3 trillion over 2025–2030, with industry-wide compute/data-center/power spending expected at a baseline $7.6 trillion between 2026 and 2031. This is the largest single-industry capital gamble in human commercial history.

Table 5: Who Actually Makes Money in This Chain?

Source: company 2025–2026 filings and analyst estimates.

The one-sentence takeaway: every dollar raised by downstream model companies ends up, unchanged, in Nvidia’s checking account.

3. Conclusion: This Model Cannot Continue

When the vast majority of downstream companies in an industry are burning $1.60 for every $1 they earn, kept alive by the next round of capital injection, and the only steady winner is the upstream hardware seller—that structure has never once held up in business history.

Venture capital is not infinite. Wall Street and sovereign funds are already demanding that these companies produce genuine, industrial-grade, profitable products within 1 to 2 years—or funding channels will be shut.

When patience runs out, if this industry still cannot deliver a true industrial-grade product, the bubble will burst faster than it inflated.

Part Two: Unable to Industrialize — The 85% Accuracy Death Swamp Is Choking Everything

Burning money is not the deadliest problem. The deadliest problem is that these expensively-burned models have not reached true industrial-product grade. This is the root reason downstream enterprises are exiting en masse—and why the entire AI enterprise cannot scale.

1. 85% Accuracy: The Wall Every Real User Hits

Anyone who has actually used ChatGPT, Claude, or Gemini on real business tasks runs headlong into the same wall:

On microscopic details, hard logic, data verification, and closed-loop code execution, these models are riddled with holes, error out at random, and confidently fabricate lies.

The simplest test: ask an AI where one of your articles has been cited. It will hand you a plausible-sounding title, author, journal, and even a fabricated URL. You check. None of it exists. Academics call this “hallucination.” In business and law, this is outright deception.

If a system cannot manage the most basic “either it exists or it does not” verification, how could it possibly be trusted with a ten-million-line national security system, a bank settlement engine, or a power-grid controller?

The technical truth is brutal: today’s large language models are, at heart, probability-based text-completion engines, not logic engines. They have no genuine causal reasoning, no common sense, no self-correction. Patch a mistake at point A today, and the model will invent a stranger mistake at point B tomorrow. Under the current Transformer paradigm, no matter how large the parameter count or how vast the compute, the accuracy ceiling may forever hover between 85% and 90%.

2. Why 85% Equals Zero — or Worse

Table 6: Commercial Viability by Accuracy Rate

Source: ISO 9001, IEC 61508 industrial quality standards; industry synthesis.

Why 85% equals zero:

• 85% means 15 random errors per 100 operations, and you cannot predict which 15;
• Companies must hire a higher-paid human to audit every output line by line;
• Checking someone else’s mistakes is often more painful than doing the work from scratch. This “pseudo-automation” does not save cost—it raises overhead and psychological burden.

Table 7: 85%-Accuracy AI vs a Junior Intern on the Same Job

3. The Enterprise Exodus: Cold Numbers

Table 8: Enterprise AI Failure Rates Across Independent Studies

Source: MIT NANDA “State of AI in Business 2025”, S&P Global, Gartner July 2026 CX report, Deloitte Tech Trends 2026, IDC 2026 mid-year. Three independent measurements converged on the same 85–95% failure band within a single year.

Table 9: Why Enterprises Are Exiting — Root Cause Attribution

Source: S&P Global 200+ enterprise study + MIT NANDA synthesis, 2026.

In one sentence: Enterprises tasted the crab. Two years in, they can neither swallow nor spit it out.

4. Why Does the Industry Stack Parameters Instead of Polishing Products?

Normal business logic would say: if a product’s accuracy is only 85%, stop and take it to 99.9% before launching. Why has the entire industry done the opposite?

Because the capital model forces them to:

• Polishing details is dull, expensive, and yields no short-term explosive returns;
• Capital markets want “Wow”—the next milestone on the road to AGI;
• Whoever stops first falls out of the funding narrative, and their valuation halves overnight;
• So all the giants sprint blindly together, using each new “bigger model” to survive one more round of storytelling.

The result is a darkly ironic, guaranteed-to-collapse loop:

1. Mercenary scientists brought in on headline contracts push architecture, never polish products;
2. Models remain stuck at 85% accuracy, unable to land in serious industries;
3. Because they can’t land, revenue can’t cover cost—so the companies survive on capital;
4. But capital only continues if there’s a next “wow” model to tell a story about;
5. So they keep poaching, keep burning cash, keep announcing new models—back to step 1.

5. The Nvidia Counter-Example: Only Companies That Perfect the Foundation Deserve to Earn

Table 10: Two Diametrically Opposite Technology Paths

• In 2006, when Nvidia launched CUDA, GPUs were still gaming toys. They poured a billion-plus dollars a year into a platform no one used. Their stock dropped 70%. Wall Street mocked them as delusional dreamers.
• They did twenty years of brutal, patient work: no flashy launches, no AGI mythology. Just one guarantee—“my chips compute faster, more accurately, and never crash.” They deliver 100% determinism—running for months on end without burning out or misplacing a decimal.
• The outcome: every AI scientist alive today learned computational programming on Nvidia tools, permanently bound to the CUDA ecosystem.

This is the wisdom of the shovel seller: most gold-diggers die on the road because their products are broken and unprofitable, but the person selling them water and sturdy shovels along the way always wins.

Any technology that cannot stand firmly on the microscopic details will collapse in the macroscopic market test. The steam engine, electricity, and the computer launched industrial revolutions only because, the day they left the laboratory, they had already reached 99%+ stability. A steam engine with a 15% chance of exploding would never have launched anything.

Conclusion: The Two Deadlocks Lock Each Other, and the Industry Cannot Continue This Way

Back to the two original problems:

One: this model cannot generate profits. OpenAI’s 2025 operating loss was $20.9B and its 2026 loss is projected to widen to $14B in cash terms (up to $25B GAAP); the four hyperscalers will together spend $725B on capex in 2026, a 77% year-over-year jump, and none has recovered its costs. Among downstream model labs, only Anthropic reached break-even in Q2 2026 ($559M operating profit)—and only through deep compute subsidies. The only companies steadily profiting are the upstream shovel-sellers Nvidia and TSMC, netting nearly $200B a year.

Two: this model cannot industrialize. Because the industry refuses to do the foundational polishing work, 85% accuracy is a ceiling no model can break through—MIT shows 95% of enterprise AI projects failed to produce measurable business impact, S&P Global shows 46% of projects were killed before production, Gartner shows 85% of AI CX systems are being dismantled, 48% of CIOs openly admit “massive disappointment.” No model has reached true product grade, so it cannot be scaled into serious industries, so real commercial closure never forms, so AI’s real utility cannot be released.

And the two deadlocks lock each other:

• No industrialization → revenue can’t cover cost → no profit;
• No profit → must rely on capital transfusions → capital demands new-model stories → keep stacking parameters, keep skipping details → no industrialization.

This loop cannot go on forever. Capital is not infinite; landlords have no infinite grain. When more and more enterprises discover that “AI integration” only means ballooning management costs and endless bugs, and when more and more investors realize that their tens of billions bought only a “loudly hyped but glitch-ridden chatbox”—panic and capital flight will erupt in a single wave.

There is only one way out: stop, and take one thing to 99.9%.

Not ten things to 85%. Not the next larger model. But in a single vertical—pure automated accounting, absolutely reliable junior code generation, or perfect intelligent customer service—patiently lift accuracy from 85% to 99.9%. This is grinding work, dirty work, the kind of work capital markets hate—but it is the only path this industry can survive on.

When the tide of capital recedes, those who actually endure and earn real money will not be the giants announcing new models every quarter. They will be the “boring” companies willing to sit on a cold bench for twenty years and perfect one specific function into industrial-grade certainty—just like Nvidia today.

The arrogance of trying to fly before learning to walk is destined for judgment by common sense and the iron laws of business. This is not our opinion; this is a rule that every chapter of business history repeats.

还没学会走就想飞：数据揭示当前 AI 产业的两个死结

作者：殷彤博士（Dr. Tong Yin） · InsightBridge Global LLC 创始人兼首席执行官

引言

在 2026 年的硅谷发布会上，Meta、OpenAI、Google 的高管们轮番宣告——通用人工智能（AGI）即将改写一切，AI 会全面取代劳动力。华尔街买账，媒体跟进，估值一路狂飙。

然而，只要拨开这些天价”雇佣兵科学家”和无限资本堆砌起来的烟雾弹，直面地面上的商业现实，你会得出一个截然相反的硬核结论：当前的 AI 产业，正陷入一场”还没学会走，就盲目想飞”的畸形怪圈。

这不是情绪化判断。它可以用两组冰冷的事实来验证：

• 第一，这套发展模式根本无法盈利——除极个别公司外，几乎所有下游 AI 巨头都在天文级亏损，而 2026 年四大云的 AI 资本开支同比暴增 77%。
• 第二，因为不愿意做基础性的苦功，没有一个模型达到真正的工业产品级别——85% 的准确率意味着无法在任何严肃行业规模化推广，MIT 最新研究显示 95% 的企业 AI 试点分文未回本。

以下用数据和事实，把这两个死结一一拆开。

第一部分：无法盈利——一场靠资本输血苦撑的巨额亏损游戏

要看懂这场泡沫是怎么烧钱的，先看谁在推动它。

一、“雇佣兵科学家”：1 亿美元合同背后的空转游戏

为了在 AI 军备竞赛中追赶 OpenAI，Meta 已经开出了 1 亿美元级别的薪酬包去挖角顶级华人科学家。乍看是天价，拆开却是这样的结构：

表 1：1 亿美元薪酬包的真实拆解

来源：硅谷标准薪酬结构（Base + Signing + RSU），按 4 年均匀归属测算。

而这个圈子的真实数据是：

• 顶级 AI 科学家在一家企业的平均在职时间只有 1.5 到 2.5 年；
• OpenAI 员工的中位数在职时间不过 16 到 18 个月（Business Insider）；
• 意味着这些天价合同从来没有被真正兑现完整。

一个科学家在 Meta 干 1.5 年拿走大约 5,400 万美元，剩下 4,000 多万未成熟股票直接作废——然后转身跳到下一家巨头，下家再用一笔更夸张的”买断签字费”把这部分损失一次性补上。他们像科技界的顶级雇佣兵，两年一跳，永远在中途兑现。

这套机制对科学家个人是天大的好事，对行业却是灾难：

• 研发”快餐化”：一个千亿参数大模型从立项到发布通常只需 6 到 12 个月，科学家追求的是发论文、跑分、突破榜单，而不是把产品打磨到可用。
• 技术断代：前任科学家刚走，继任者思路不同，代码往往推倒重来，几十亿美元的算力就此蒸发。
• 只管上天，不管落地：模型架构一搭好，日常维护、边界处理、稳定性优化——这些真正产品化需要的脏活累活，没有人愿意干，也没有人有时间干。

二、下游模型公司的真实盈亏账本

在天价挖角、天量算力的持续烧钱下，2025–2026 年下游 AI 企业的真实盈亏是这样的：

表 2：OpenAI 2024 vs 2025 财务对比（审计文件泄露版）

来源：Ed Zitron / Financial Times / Fortune / Ars Technica 综合报道，2026 年 6 月。注：2025 年 385.3 亿美元净亏损含 415.5 亿美元一次性非现金准备（非营利转营利结构调整所致），剥离后的可比现金亏损约 80 亿美元；运营亏损 209.2 亿美元为最能反映真实经营的口径。

表 3：全球顶级 AI 实验室 2025–2026 盈亏对比

来源：Fortune、Financial Times、CNBC、Mental Momentum Research 综合报道，2026 年 6–7 月。

表 4：四大云 AI 资本开支（Capex）军备竞赛

来源：Goldman Sachs、S&P Global、Yahoo Finance、Gate.com 综合报道，2026 年 6–7 月。CreditSights 测算：2026 年 hyperscaler capex 中约 4,500 亿美元直接投向 AI 基础设施。

Goldman Sachs 更进一步预测：四大云 2025–2030 累计 Capex 将达 5.3 万亿美元，2026–2031 全行业算力、数据中心、电力累计投入基线为 7.6 万亿美元。这是一个人类商业史上前所未有的资本豪赌规模。

表 5：谁在这条链上真正赚钱？

来源：各公司 2025–2026 财报、行业分析师综合估算。

一句话看懂这张表：整个 AI 链条里，下游模型公司融来的每一美元，最后原封不动地流进英伟达的支票账户。

三、结论：这套模式没办法持续

一个行业如果绝大多数下游公司的商业模式是”每赚 1 美元就烧掉 1.6 美元”，靠资本市场的下一轮输血续命，唯一稳赚的是最上游的硬件卖铲人——那么在商业史上，这种结构没有一次能撑得下去。

资本的钱不是无限的。华尔街和主权基金已经开始要求这些公司在 1–2 年内拿出真正的、能够工业级落地的盈利证明，否则将彻底关闭融资通道。

当资本的耐心耗尽的那一天，如果这个行业还是拿不出一个真正的工业级产品，泡沫破裂的速度会比它膨胀的速度还要快。

第二部分：无法工业化——85% 准确率的死亡泥潭正在阻断整个产业的推广

烧钱本身还不是最致命的问题。真正致命的是：这些烧出来的模型，没有一个达到了真正的工业产品级别。这才是阻断整个 AI 事业推广、让下游企业集体退场的根本原因。

一、85% 准确率：任何一个亲身测试的人都会撞上的铁墙

任何用 ChatGPT、Claude、Gemini 处理过真实业务的人，都会遇到同一堵墙：

在微观细节、硬核逻辑、数据查证、代码闭环上，这些模型漏洞百出、随机出错、甚至自信地编造谎言。

最简单的一个测试：让 AI 告诉你，你的某篇文章在哪里被引用发表过。它会言之凿凿地给出篇名、作者、期刊，甚至伪造出链接。你自己动手去查，根本没这回事。这在学术上叫”幻觉（Hallucination），在商业和法律语境里，这就是明目张胆的撒谎。

一个连”有就是有、没有就是没有”的单点查证都做不好的系统，怎么可能承载千万级代码的国家级安全系统、银行结算系统、电网调度系统？

技术真相很残酷：目前的大模型本质上是一台”概率接龙机器”，不是逻辑机器。 它没有真正的因果推导、没有常识、没有自我纠错能力。你今天微调修好 A 处的错，明天它会在 B 处以一种更诡异的方式重新犯错。顶级科学家们心里清楚：在现有的 Transformer 技术路线下，无论怎么堆参数、砸算力，正确率的天花板可能永远卡在 85% 到 90% 之间。

二、为什么 85% 就意味着”零分”甚至”负分”

在几乎任何真正的商业场景里，85% 的准确率都远远不够——甚至比零分更糟。

表 6：不同准确率下的商业可用性

来源：ISO 9001、IEC 61508 工业质量标准；行业分析综合。

为什么 85% 是零分？

• 85% 意味着每 100 次操作中有 15 次是随机错误，且无法预知在哪里出错；
• 企业必须安排一个薪水更高的高级员工去逐行”人肉排雷”；
• 检查别人的错漏，往往比自己从头做还痛苦。这种”伪自动化”不但没有降本增效，反而增加了管理成本和心理负担。

表 7：AI 与人类员工在同一岗位的对比

三、企业退场潮：冷酷的数字

以下是 2025–2026 年多份权威机构追踪数据的集中汇总：

表 8：企业 AI 落地的失败率数据

来源：MIT NANDA “State of AI in Business 2025”、S&P Global、Gartner 2026 年 7 月客服 AI 报告、Deloitte Tech Trends 2026、IDC 2026 中期报告。三份独立测量在一年内向同一个数字（85%–95% 失败率）收敛。

表 9：企业为什么退场——原因归因

来源：S&P Global 200+ 企业调研 + MIT NANDA 报告综合归因，2026 年。

一句话：大企业下场吃螃蟹，吃了两年，现在吃不下也吐不出来。

四、为什么整个行业选择了”堆参数”而不是”打磨产品”？

正常的商业逻辑应该是：如果一款产品的准确率只有 85%，就应该停下来把它做到 99.9% 再推广。为什么整个行业反着来？

因为资本模式逼着他们必须这样做：

• 打磨细节是极度枯燥、投入巨大、短期看不到暴利的苦差事；
• 而资本市场要的是”震撼（Wow）“、要的是”通往 AGI 的下一个里程碑”；
• 谁先停下来打磨产品，谁就在融资故事里出局，估值立刻腰斩；
• 于是所有巨头只能一起蒙眼狂奔，用一个又一个”更大的新模型”来续命、来讲下一轮故事。

这就形成了一个极其讽刺、注定崩盘的死循环：

1. 天价合同挖来的雇佣兵科学家，只做架构突破，不做产品打磨；
2. 模型永远卡在 85% 的准确率，无法在严肃行业落地；
3. 因为落地不了，收入远远覆盖不了成本，公司只能靠融资输血；
4. 融资又必须靠”下一个震撼的新模型”来讲故事；
5. 于是继续挖角、继续烧钱、继续推新模型——回到第 1 步。

五、英伟达的反面样本：把地基做到极致的公司才配赚钱

一个刺眼的对比：为什么在这场狂欢里，只有英伟达能稳赚不赔？

表 10：两种技术路径的残酷对比

因为英伟达走的是完全相反的道路：

• 2006 年，英伟达推出 CUDA 平台时，显卡还只是打游戏的玩具。他们每年砸十几亿美元去搞一个没人用的底层软件，股价跌了 70%，被华尔街嘲笑为不切实际的疯子。
• 他们扎扎实实干了二十年苦工：不做惊天动地的发布会，不吹嘘 AGI 神话，只保证一件事——“我的芯片算得最快、最准、绝不崩溃”。输出 100% 的确定性，可以连续运行几个月不烧毁、不算错一个小数点。
• 结果：全球所有 AI 科学家从学写第一行复杂计算代码开始，用的就是英伟达的工具——早已被它深度绑定。

这就是”卖铲人”的智慧：淘金的人可能一大半死在路上（因为产品稀烂、没有利润），但一路上给他们卖水、卖坚固铲子的人，永远稳赚不赔。

任何无法在微观细节上做到脚踏实地的技术，最终都会在宏观的市场检验中轰然倒塌。 蒸汽机、电力、计算机之所以能开启工业革命，是因为它们走出实验室的那一刻，就已经达到了 99% 以上的稳定度。一台有 15% 概率会突然爆炸的蒸汽机，永远不会真正开启任何工业革命。

结语：两个死结互相锁死，产业无法继续这样走下去

回到最初的两个问题：

第一，这套发展模式没办法盈利。 OpenAI 2025 年单年运营亏损 209 亿美元，2026 年将扩大至 140 亿美元现金亏损（GAAP 口径达 250 亿美元）；四大云 2026 年 AI Capex 总计将高达 7,250 亿美元、同比暴增 77%，至今没有一家能收回成本。整个下游行业只有 Anthropic 一家在 2026 Q2 首次实现运营盈利（5.59 亿美元），且深度依赖巨头补贴。真正稳赚的只有上游卖铲子的英伟达、台积电，一年拿走近 2,000 亿美元净利润。

第二，这套发展模式没办法工业化。 因为不愿意做基础性的打磨苦工，85% 的准确率成了所有模型无法翻越的天花板——MIT 研究显示 95% 的企业 AI 试点没能转化为任何商业利润，S&P Global 显示 46% 的项目在投产前被砍掉，Gartner 显示 85% 的企业 AI 客服正在被大面积拆除，48% 的 CIO 公开承认”巨大失望”。没有一个模型达到了真正的产品级别，就没办法在严肃行业推广，就形不成真正的商业化闭环，就无法把 AI 的作用发挥出来。

而这两个死结是互相锁死的：

• 因为无法工业化 → 收入远远覆盖不了成本 → 无法盈利；
• 因为无法盈利 → 必须靠融资输血 → 融资必须靠新模型故事 → 只能继续堆参数、不打磨细节 → 继续无法工业化。

这个死循环没办法一直走下去。资本的钱不是无限的，地主家也没有余粮。当越来越多的企业发现”集成 AI”带来的只有暴增的管理成本和无穷无尽的漏洞时，当越来越多的投资人发现自己的几百亿美元只换来了一个”吹得震天响、但漏洞百出”的聊天框时——恐慌和撤资潮就会瞬间爆发。

出路只有一条：停下来，把一件事做到 99.9%。

不是十件事做到 85%，不是发布下一个更大的模型，而是在某一个具体的垂直领域——纯粹的自动化会计、绝对准确的初级代码生成、完美的智能客服——把准确率老老实实地从 85% 抬到 99.9%。这是苦工、是脏活、是资本市场不喜欢的事，但这是这个产业活下去、真正做大的唯一路径。

当资本的潮水退去，真正能留下来赚到钱的，绝对不是天天发布新模型的巨头，而是那些愿意坐二十年冷板凳、把某一个具体功能做到工业级确定性的”笨公司”——就像今天的英伟达。

还没学会走就想飞的傲慢，注定要被常识和商业规律无情审判。 这不是我们的判断，这是每一段商业史都在反复演绎的铁律。