A single New York Times report — that OpenAI is considering pushing its IPO to 2027 rather than going public this year — was enough to send the Nasdaq to five consecutive losing sessions, its worst streak since February. By Friday's close, the index had shed 4.6% on the week. SoftBank Group plunged more than 12% in a single session in Tokyo. South Korea's Kospi, which had been up nearly 95% this year on the back of chip demand, tumbled 10% on Tuesday alone and briefly triggered a circuit breaker. The question that followed all of it: is the market finally asking whether the AI buildout is built on something real?

What the numbers show

The damage was concentrated but global. In the US, the Nasdaq Composite closed Friday at 25,297.62, down 0.24% on the day and approximately 4.6% on the week, according to CNBC. The S&P 500 lost nearly 2% over the same five sessions, while the Dow Jones — less exposed to tech — actually rose 0.6% for the week, a divergence that illustrates exactly where the selling was concentrated. In Asia, the hit was sharper: South Korea's Kospi fell 10% in a single session on June 23, with SK Hynix and Samsung each dropping more than 12%; those two companies represent roughly half of the index's total market capitalization. SoftBank, which holds a 13% stake in OpenAI, fell more than 12% on Friday alone. European chip stocks followed suit: ASML was down 2.2%, Infineon fell 3.7%, and STMicroelectronics lost 3.3%.

Index / Stock Move (week / session) Context
Nasdaq Composite−4.6% (week)Longest losing streak since February 2026
S&P 500−~2% (week)Closed at 7,354.02 on Friday
SoftBank Group−12%+ (Friday)Holds 13% stake in OpenAI
Kospi (South Korea)−10% (June 23)Circuit breaker triggered; SK Hynix & Samsung −12%+
ASML (Netherlands)−2.2% (Friday)European chip sector broadly lower
Dow Jones+0.6% (week)Defensive rotation; less tech exposure
Watch out: JPMorgan traders noted in a note that the OpenAI IPO delay raised concerns about the "sustainability of infrastructure spending given the delay in funding from the capital markets." That's the structural worry under the surface — not OpenAI's revenues, but whether the entire AI capex cycle can hold without a mega-IPO to recycle capital back into the system.

Why this matters for freelancers

If you work in tech — as a developer, AI consultant, data analyst, or any adjacent role — this week is worth paying attention to. Not because the market determines your next invoice, but because the AI spending cycle does. The hyperscalers (Microsoft, Amazon, Google, Meta, Oracle) have committed somewhere between $600 billion and $725 billion in capital expenditure for 2026, the overwhelming majority directed at AI infrastructure: data centers, GPUs, and the compute needed to train and run large models. That spending is what's driving demand for AI skills, AI tools, and AI-adjacent freelance work. If the market's skepticism about ROI on that spending starts to translate into actual capex cuts — which hasn't happened yet — the demand side for tech freelancers changes. For now, the signals are mixed: cloud revenues are growing, enterprise AI adoption is broadening, and none of the hyperscalers have announced pullbacks. But the tone has shifted from "spend first, ask questions later" to something more cautious, and that's worth tracking.

Context: how we got here

OpenAI confidentially filed its S-1 with the SEC on June 8. The company had been widely expected to go public in 2026, potentially in the third or fourth quarter, at a valuation approaching $1 trillion. By March 2026 it had raised $122 billion in a single round — the largest in private company history — and was generating roughly $2 billion per month in revenue, with an annualized run rate above $25 billion according to Reuters and Sacra. The problem is that OpenAI is also projected to lose approximately $14 billion in 2026 on that same revenue, and its cash burn is expected to be around $27 billion according to research firm Sacra. When SpaceX went public and its stock quickly retreated from its post-IPO high, advisers reportedly told Sam Altman that the retail market might not be ready to absorb a trillion-dollar listing right now. Altman's reported response: a lower valuation was a non-starter. So the choice became wait for 2027 or price below $1 trillion — and the market reacted as if the delay itself was the news. It wasn't, exactly. What it revealed was how much of the AI trade had been priced around the expectation of a near-term liquidity event, not around the underlying business.

$725 billion in AI capex committed for 2026 — and the biggest potential IPO of the era just blinked.
Calcugui analysis · June 2026

The South Korea sell-off had additional local catalysts — MSCI again excluded the country from its Developed Markets watchlist, and regulators raised concerns about leveraged single-stock ETFs on Samsung and SK Hynix that had just launched the prior month. But the broader AI repricing was the accelerant. The Kospi had run nearly 95% this year on the back of memory chip demand tied to AI; that kind of move doesn't unwind gently when sentiment shifts.

What comes next

The IPO itself hasn't been cancelled. OpenAI filed confidentially, which gives it the flexibility to go public if conditions improve — and Kalshi prediction markets currently put a 59% chance on an official announcement by March 1, 2027, and 73% by June 2027. The key variable isn't the IPO; it's whether the hyperscalers' Q2 earnings (due in July) confirm that AI revenue is actually scaling alongside capex. If Microsoft's Azure AI growth holds above 60% year-over-year, if Amazon's AWS data shows enterprise adoption broadening, that's the data that stabilizes the trade. If guidance disappoints, the conversation about a bubble gets louder. The Nasdaq is still up roughly 10% year-to-date despite this week's drawdown — it's not a collapse, it's a gut-check, as Wedbush's Dan Ives put it. But gut-checks in overextended trades can become something more if the fundamentals don't hold.

Key insight: The AI trade is splitting in two. On one side: companies that can show actual AI-driven revenue growth (hyperscalers with strong cloud numbers, HBM memory suppliers like Micron and SK Hynix with real demand). On the other: pure narrative plays that have been priced as if the infrastructure cycle is a guarantee. The market is now asking for receipts. That's not a bubble pop — it's a repricing. The distinction matters for how you read it as a freelancer in tech.
📡 Signals to watch
🔴
Hyperscaler Q2 earnings — July 2026 Microsoft, Amazon, Alphabet, and Meta report next month. Azure AI revenue growth, AWS backlog data, and capex guidance will either confirm the buildout thesis or accelerate the repricing. This is the most important data point in the near term.
🟡
OpenAI S-1 timeline — will the company go public in 2026? The confidential S-1 filed June 8 keeps the option open. Watch for any news of pre-IPO investor roadshows or SEC filing updates. A public S-1 would signal a 2026 debut; continued silence likely means 2027.
🟢
Micron and SK Hynix — HBM memory demand Both companies are direct suppliers of the high-bandwidth memory that AI chips require. If their order books hold and guidance stays strong, it's evidence that the AI compute buildout is real demand, not inventory speculation.

Know exactly what you're making — before taxes, fees, and FX.

Calcugui runs the numbers on your freelance income so market noise doesn't catch you off guard.

Try the Calculator →
📩 Weekly newsletter
Get the weekly freelance income breakdown — free
No spam. Just what matters for your income as a freelancer, every week.

📚 Recommended courses
Artificial Intelligence A-Z 2026
⭐ Bestseller · Udemy
Artificial Intelligence A-Z 2026
Build 8 real AIs with LLMs, Deep Learning and Agentic AI. 200K+ students. From USD 14.99.
View course on Udemy →
Browse all recommended courses →
💼
Freelance Jobs Board
Looking for work? Find freelance listings on Calcugui →