High Rates, Higher Stakes for AI

Author

Nicole Rojas
October 20, 2026

 

Welcome to AlphaInsights, 8alpha.ai’s weekly newsletter, your ultimate source for curated insights and key updates from the dynamic world of venture capital!

From billion-dollar rounds to market-defining shifts, we deliver the intelligence powering the global investment landscape, moving investors and innovators forward. At 8alpha.ai, we’re not waiting for the future of capital, we’re building it. Stay sharp, stay curious, and stay ahead.

STARTUPS

ROUNDS AND UNICORNS

  1. Modal Labs (AI Infrastructure): New York-based Modal Labs, a cloud infrastructure provider for AI inference, secured $750 million in a Series D led by Accel at a $15.75 billion valuation, and says it recently passed a $1 billion annual revenue run rate

  2. Icon (Construction Tech): Austin-based Icon, a developer of robotic construction systems known for 3D-printed homes, closed $200 million in Series D funding led by Baillie Gifford and BlackRock, the company is on pace for 4x revenue growth in 2026

  3. Soufflé Therapeutics (RNA Therapies): Boston-based Soufflé Therapeutics, an RNA therapies developer, raised $135 million in Series B funding led by Juvia Group as part of a large investor syndicate

  4. Estuary Biotherapeutics (Gene Therapies): Menlo Park-based Estuary Biotherapeutics, which develops genetic medicines for serious kidney diseases, launched with $103 million, including a $93 million Series A led by RA Capital Management and $10 million in previously undisclosed seed capital

  5. Labelbox (AI Tools): San Francisco-based Labelbox, which builds technology to train and evaluate AI models and deploy agents that improve with experience, picked up $100 million in Series E funding led by Infinitum Partners

 

Staying at the AI frontier for five years costs about $100 billion according to PitchBook’s model: roughly $93 billion in cash spending plus a $7 billion reserve, with outside financing peaking near $29 billion. Capital is concentrated: in H1 2026, horizontal platforms took $288.1 billion, or 70.8% of global AI venture value, while vertical applications made up 62.9% of the 3,500 deals, and five rounds supplied 62.3% of all AI funding. Research beyond the final training run is the biggest swing factor, with the budget ranging from $106.5 billion to $142.5 billion depending on how much research compute it takes, and customer cash arrives late, with 60% of the $80 billion collected over five years landing in year five

  • Anthropic reportedly expects to spend at least $518 billion on infrastructure over roughly a decade, about 80% of it noncancelable

  • PitchBook’s example shows how dilution works against investors: $10 million at a $1 billion valuation, diluted 20% in each of two later rounds, ends at 0.64% ownership and needs an exit above $1.56 billion just to break even

  • CoreWeave's $1.51 billion of Q2 adjusted EBITDA became a $626 million net loss after depreciation and interest expense

Latin American startups raised $1.3 billion across seed and growth deals in Q3 2026, up 26% year over year but down 9% from Q2. Late-stage and growth funding totaled $757 million, up 74% from a year earlier but down 21.3% from Q2, while angel and seed funding surged to $266 million, up 128% from Q2, led by Brazil's Decade, whose $85 million round was the largest seed deal ever raised by a Latin American startup. Mexico has raised $1.52 billion in the first three quarters, about 40.7% more than Brazil's $1.08 billion, though Mexico's total is concentrated in a few startups while Brazil has more deals and leads in realized exits

  • Mexico's Kapital raised a $125 million Series C at a $1.6 billion valuation

  • International VCs were more active in the region in the first three quarters than in 2024 and 2025 combined

  • About 60 private Latin American tech companies have each raised at least $150 million, a potential IPO pipeline, though exits are coming mostly through acquisitions and secondaries

ECONOMIC SNAPSHOT

Prices rose 0.6% in September, pushing annual inflation to 3.7% from 3.4% in August and above the 3.6% economists expected, with pressures extending well beyond gas. Gas prices rose 9.1% and drove more than half of the monthly increase, but core CPI also ran hot at 0.3% for the month and 2.6% year over year, versus expectations of 0.2% and 2.5%, with core goods prices rising at their fastest rate since January 2025 and shelter up 0.4%. With wages up about 3%, the gap between inflation and pay is widening. Stocks fell and the 10-year Treasury yield jumped eight basis points to 5.3%, while the odds of an October Fed rate hike rose to 49% from 22% a day earlier

  • Grocery prices fell 0.1% from August, though higher diesel costs could push food prices up

  • Computer software and hardware prices are rising at their fastest-ever annual rates, reflecting AI-related infrastructure spending

  • Inflation is expected to stay elevated through the end of the year

Fed Chair Kevin Warsh warned that surging AI spending is a "shock" to the US economy and pledged to contain price increases as a flood of private funds drives up demand, speaking at an IMF conference in Bangkok. Investment in chip manufacturing and data centers is hitting the demand side of the economy before it adds supply, which could push consumer prices higher. Warsh remains optimistic that AI will ultimately raise productivity and wealth, but called the path there "very bumpy," with uncertainty over market structure and "who's going to be making the money?" The AI shock comes on top of the Iran war's oil supply shock, with inflation at 3.7% in September, more than a percentage point above the Fed's 2% target

  • The next Fed meeting is Oct. 28, days before the midterm elections, and Warsh declined to say what the Fed might do

  • Investors now put the odds of the Fed holding rates at about 86%, up from about 55% on Wednesday, per CME FedWatch

  • Rate hikes typically take several months to slow demand and bring down inflation

 

Is the US economy as narrow as its stock market? (Financial Times, 4 minute read)

The US economy is less dependent on AI than the stock market is, with one asterisk. Consumer spending, not AI, is still the main engine, contributing 2.5 percentage points to Q2 real GDP growth, almost three times all private investment. AI data center construction runs at $85 billion a year, about a quarter of a percent of GDP, and one Fed estimate puts the AI boom's contribution to growth at about 0.6 percentage points a year. The asterisk is the wealth effect: household shareholdings have grown by $34 trillion since ChatGPT launched, equities are a record 33% of household assets, and a Bank of France model attributes about 40% of 2025 consumption growth to it. A bursting AI bubble could tip the economy into recession, making this a dependence on AI euphoria rather than AI investment

  • The share of sectors growing in value added and employment is in line with its 20-year average

  • About 86% of spending categories are expanding, per Bank of America data, with growth across income groups

  • Pimco's Dan Ivascyn said the 10-year Treasury yield could reach 6%, up from 1.5% five years ago

 

Tech stocks had a rough week as higher interest rates and mounting AI debt weighed on sentiment. The Nasdaq fell 4.5% on the week and the S&P 500 almost 2%, both their worst weeks since June, just 10 days after record highs. At the center is CoreWeave, whose tens of billions of dollars of debt has pushed its bond yields to distress levels. It lost a quarter of its value this week and is down more than 65% from its March 2025 post-IPO peak. A semiconductor index fell 11.4% on the week, its eighth straight daily decline, after Chinese memory chipmaker CXMT announced progress on advanced memory chips, and South Korea's Kospi tumbled 14.5%, its worst week since 2008. The VIX jumped more than 20% to above 20

  • The semiconductor index is down more than 20% from its June peak but still up more than 50% year to date

  • Semiconductors account for roughly one-fifth of the S&P 500's market value, so further losses could hurt the broader market

  • The S&P 500 and Nasdaq are down roughly 2% and 5%, respectively, from their peaks earlier this month

 

Cracks are appearing in America's debt-laden $10 trillion AI infrastructure boom, with at least 47 large data center projects delayed, withdrawn, or blocked in 2026 and Bernstein estimating only about 35% of the planned US pipeline is credible. More than 1,000 US communities have tried to restrict AI infrastructure this year, and higher rates and a souring bond market threaten the neoclouds that must finance construction. A CoreWeave bond fell into distress, SoftBank’s SB Energy needs $174 billion to deliver $430 billion in contracts and has delayed its IPO, and Nscale has just 25,000 of a planned 500,000 chips online. US AI infrastructure spending is projected to top $10 trillion from 2025 to 2032, about 3.6% of GDP a year

  • Nebius's CEO said its planned 5 gigawatts of capacity would take about $250 billion of financing, with only about 20% raised so far

  • Cipher agreed to absorb the first $359 million of cost overruns on its Texas lease with Fluidstack, requiring about $750 million of additional equity

IPOs & EXITS

IT has become a major driver of GP-led secondaries, accounting for 24.6% of global GP-led secondary exit value in 2025, up from 10.6% in 2023, as sponsors that can't sell IT businesses roll them into continuation vehicles. IT's share of GP-led deal count rose to 19.7% from 13.4%, and in single-asset continuation vehicles, IT deal value climbed from $1.9 billion in 2023 to $7.3 billion in 2025 while deal count doubled from five to 10. The shift reflects slower IT exits since the 2022 rate hikes and the AI-driven software selloff, with firms that bought at peak 2021 multiples struggling to sell at today's lower valuations and using continuation funds to buy time to reposition assets

  • Bregal Unternehmerkapital closed an €811 million (about $909 million) multi-asset continuation fund in July, co-led by LGT Capital Partners and Pantheon

  • The fund rolls over Safety21, a Rome-based GovTech platform, and Onlineprinters, a German online printing services provider

Private equity funds are increasingly pushing portfolio companies to take on more debt to return cash to investors, as IPOs and sales to other buyers have dried up. "Dividend recap" activity has surged since 2022, and Moody's estimates PE-owned companies in Europe, the Middle East and Africa could add another €41 billion ($46 billion) of debt before reaching leverage levels that typically prompt a downgrade. Distributions to investors have fallen from more than 30% of starting net asset value in 2022 to about 12% today, with funds holding companies well beyond the typical five to six years. The risks are that recaps leave companies with less room to invest or weather downturns, and that a prolonged exit backlog could push sponsors to monetize weaker credits

  • Moody's analyst Mayank Mallik said exits are "few and far between," and that debt documentation is more permissive than a decade ago

  • Raymond James' Sunaina Sinha Haldea said recaps give liquidity but don't resolve the valuation gap, since the asset stays in the fund and still needs a buyer

  • Services, consumer products, and manufacturing are the most active sectors, with most recaps going to companies that beat their business plans

 

Mega-IPOs broaden appeal of tycoon wealth tactics (Reuters Breakingviews, 6 minute read)

Mega IPOs like Anthropic's may create a new class of everyday millionaires, but employees often can't tap that wealth before their company lists, and selling shares sacrifices upside and triggers taxes. Most startup employees don't work at unicorns: the median employee is more likely to see a low five-figure payout, with their firm selling for under $100 million. A backlog of giant private companies worth $5.3 trillion can't or won't go public, and the median age of IPO companies has risen to 14 years. Companies are responding with tender offers, which on Nasdaq Private Market rose from about $3 billion in 2023 to just under $15 billion in 2025, while borrowing against private stock, long reserved for the ultra wealthy, is becoming another option, with Secfi alone facilitating about $1 billion in financing

  • Private startup employees can have up to 86% of their net worth in company stock, versus just over a third of US household net worth for stocks overall

  • OpenAI recently bought back $7 billion of employee shares in a tender offer

  • Employees typically own more stock than founders by the Series C stage, which widens the pool of potential borrowers

Newly public crypto companies have mostly performed poorly despite a booming stock market, with two exceptions: Circle and Hyperliquid Strategies, which is up over 200%. Among traditional IPOs, only Circle trades near its listing price, thanks to its outsized opening-day pop, while Figure is down about 19% against a 13% S&P 500 return over the same period. BitGo, Bullish, and eToro are down 55% to 65%, and Gemini has fallen nearly 90%, shrinking its market cap from $3.3 billion to about $500 million. Among non-traditional listings, American Bitcoin is down over 93% and CoinShares over 50%. Frothy 2025 conditions let weak IPOs through, and many laggards, like Gemini, Bullish, and BitGo, offer commodity services in a hyper-competitive market where they face stronger rivals with deeper pockets

  • Payward (formerly Kraken), MetaMask, and Ledger have signaled they want to IPO soon

  • Firms with scale and a strong product, like Circle, or a strong team with a clever model, like Hyperliquid Strategies, can succeed

Prediction market traders on Kalshi see Anthropic as the only major company likely to announce an IPO this year, giving it 75% odds as of Monday (Oct. 19), the only one of 23 open contracts above 50%. Expectations have collapsed for others: Plaid fell from 50% in January to 10%, Deel from 50-50 in February to 3%, and Databricks from over 50% to 1%, while OpenAI is at 3% after Sam Altman ruled out a 2026 listing. IPO proceeds are up sharply this year on megacap debuts like SpaceX, but the number of US IPOs priced is down 33% from 2025, per Renaissance Capital, and postponements have grown. Anthropic's Q3 revenue of $16.8 billion showed slower quarter-over-quarter growth than in past periods, and its IPO valuation range now reportedly starts at $1.5 trillion versus the prior expectation of at least $2 trillion

  • Anthropic confidentially filed its prospectus with the SEC in early June, and traders give OpenAI 50-50 odds of announcing an IPO by June 2027

  • Kalshi counts an IPO as officially announced if the SEC declares the S-1 effective, the IPO prices, or an exchange assigns a ticker

  • 22V Research called the lower valuation range disappointing and said questions remain on why growth slowed in September and whether Anthropic can hit expected growth given pricing pressure on large language models

 

8ALPHA.AI HIGHLIGHT

"We must slow the pace at which we improve the capabilities of AI models."

— Dario Amodei, CEO of Anthropic, "We Must Pace the Frontier" (September 2026)

It is a strange sentence to read when the person who wrote it is reportedly preparing the largest stock-market listing in history. Dario Amodei, CEO of Anthropic, the world's fastest-growing AI company, asked the industry to slow down. Sam Altman, the CEO of OpenAI, its biggest rival, agreed. After years of racing to build increasingly powerful machines, the people leading the race are suddenly asking everyone to hit the brakes. Why?

Artificial intelligence is an economic experiment of extraordinary scale, financed by record amounts of capital, driven by a handful of companies, and unfolding faster than businesses, governments, and workers can adapt. The promise is enormous: greater productivity, accelerated economic growth, and the automation of work once considered uniquely human. So is the price tag.

Capital is concentrating into an increasingly exclusive club of companies, with valuations and funding rounds that would have seemed absurd just a few years ago. At the center of the frenzy are OpenAI and Anthropic, whose valuations have exceeded the trillion-dollar threshold. Their anticipated public listings could redefine capital markets. Paper wealth is abundant.

But the stakes extend far beyond Silicon Valley. AI could fundamentally reshape the relationship between economic growth and employment. For the first time, the productivity revolution may be less about helping people do their jobs and more about figuring out if people are still needed.

The economy and financial markets are increasingly exposed to a single outcome: AI delivering on its extraordinary promises. With 86% of US venture dollars in the first half flowing into AI and roughly half the S&P 500's market value tied to AI-linked companies, the margin for error has disappeared.

The future may indeed be intelligent, abundant, and extraordinarily profitable. The question is profitable for whom. And at what price?

8alpha.ai is an AI fintech transforming cash-generating businesses into scalable, AI-powered companies. We provide revenue-based financing and hands-on AI transformation, delivering no zeros with unlimited upside. We’re the architects building financial infrastructure for the next generation of investors and startups.

Become part of our revolution.

Happy reading,

8alpha.ai’s Research & Investment Team