Goldman Sachs Says AI Added 'Basically Zero' to GDP — And 90% of CEOs Agree

Goldman Sachs just dropped a bomb: massive AI investment contributed "basically zero" to U.S. economic growth last year.

And a new NBER study of 6,000 executives across the U.S., U.K., Germany, and Australia confirms it — nearly 90% say AI has had no measurable impact on employment or productivity over the past three years.

Let that sink in. Companies are spending $250 billion annually on AI infrastructure. And the return? Statistically invisible.

Solow's Ghost Is Laughing

In 1987, economist Robert Solow said: "You can see the computer age everywhere but in the productivity statistics."

Forty years later, we're living the exact same paradox — just with fancier models.

The NBER data is brutal:

• Two-thirds of executives use AI, but only 1.5 hours per week
• A quarter of respondents don't use AI at work at all
• Workers' AI usage went up 13% in 2025, but confidence in AI's utility dropped 18%

More usage, less trust. That's not adoption. That's disillusionment.

The Real Problem: AI Is Automating Waste

Here's the uncomfortable truth most AI companies won't tell you: if your job is 60% meetings, 30% emails, and 10% actual value creation — making the emails faster doesn't create value. It creates faster waste.

Most companies are stuck in the "do the same stuff but with AI" phase. They bolt ChatGPT onto broken workflows and wonder why nothing changes.

The productivity gains won't come from AI doing your existing tasks faster. They'll come when AI changes what work gets done.

But Wait — The Models Are Getting Insane

Here's where it gets interesting. While CEOs report zero impact, the models themselves are improving at breakneck speed.

Google just released Gemini 3.1 Pro with a 77.1% score on ARC-AGI-2 — a benchmark for novel logical reasoning. That's 2.5x better than Gemini 3 Pro from just three months ago. Claude Opus 4.6 is pushing boundaries in agentic coding. GPT-5.3 Codex is rewriting how developers ship software.

The models aren't the bottleneck. We are.

The J-Curve Bet

Some economists see a J-curve: initial slowdown, then exponential surge. The electricity analogy is popular — it took 30 years after Edison's first power station for factories to be redesigned around electric motors instead of steam.

But here's what I think separates winners from losers right now:

Winners aren't using AI to do old work faster. They're using AI to do work that was previously impossible — automated meeting transcription that feeds directly into action items, AI agents that handle entire customer support workflows end-to-end, voice cloning that lets one person produce content in 29 languages.

If you're still manually transcribing meetings or writing notes by hand, tools like Fireflies already handle this automatically — the free tier covers 800 minutes of transcription per month, which is enough for most teams to see immediate time savings.

The Developer's Edge

For developers reading this: the paradox is actually your opportunity.

While enterprises fumble with AI strategy decks and "transformation roadmaps," individual developers and small teams can move fast. The gap between what AI can do and what companies are actually doing with it is enormous.

That gap is where the money is.

If you're building AI-powered tools or workflows, the market isn't saturated — it's barely started. And if you're deploying anything that needs reliable infrastructure without enterprise pricing, Vultr's cloud platform starts at $2.50/month with GPU instances available for AI workloads.

My Take

The AI productivity boom is real. It's just not evenly distributed yet.

The 10% of companies seeing results aren't using better models — they're using the same models with better organizational design. They've restructured workflows, retrained teams, and most importantly, they've stopped trying to make AI fit into old processes.

The question isn't whether AI will transform productivity. It's whether your company will figure it out before your competitors do.

What's your experience? Is AI actually making you more productive, or are you in the 90%?

---

Data sources: NBER Executive Survey (2026), Goldman Sachs Economic Research, ManpowerGroup 2026 Global Talent Barometer, The Economist