FT : Bank of Japan raises rates to 1% for first time since 1995

Bank of Japan raises rates to 1% for first time since 1995
Central bank says it will stop reducing level of monthly bond purchases from next year

The Bank of Japan has raised its short-term policy rate to “around 1 per cent”, taking the cost of borrowing to its highest level in 31 years as the country adjusts to sustained inflation. 

The 0.25 percentage point increase, which was widely expected, takes Japan to what analysts said was a critical milestone in the central bank’s effort of normalising monetary policy after years of ultra-low interest rates and deflation.

The BoJ’s policy rate was last at 1 per cent in 1995, when the central bank was in the process of lowering borrowing costs in the wake of the Japanese asset bubble burst in the late 1980s.

In a statement accompanying the decision, the BoJ signalled that it intended to continue that normalisation process, raising the policy interest rate and degree of monetary accommodation “in response to developments in economic activity and prices as well as financial conditions”.

The BoJ also said that from April 2027 it would stop reducing its monthly purchases of Japanese government bonds, levelling off at a pace of about ¥2tn of monthly purchases. That move was also widely expected by the market.

The yen held steady at about ¥160.2 versus the dollar following the announcement.


The BoJ said that while higher crude oil prices were weighing on economic activity, “the risk of a significant slowdown in the economy appears to have decreased compared with a while ago”.

It also noted that the price pass-through from higher fuel prices had been progressing relatively quickly, and could spread from business-to-business transactions to push underlying consumer price inflation above its target of 2 per cent.

Since lifting Japan out of negative interest rates in 2024, the BoJ raised rates twice in 2025. It has been expected to settle into a pattern of gradually tightening every six months or so. Some economists believe a further 0.25 percentage point rise could come as soon as October.

This week’s decision to raise interest rates this week was reached by a 7-1 vote of the bank’s Monetary Policy Committee, which was down to eight members after governor Kazuo Ueda was admitted to hospital last week.

The dissenting member, Toichiro Asada, argued that the situation in the Middle East presented Japan with greater downside risks to production and employment than the upside risks to prices.

“The distribution of votes is interesting and reflects that the board is a bit more balanced now when previously it skewed comfortably hawkish,” said Stefan Angrick, head of Japan at Moody’s Analytics.

“The fact is also that the BoJ has no good choices,” he added. “They can hike to stem inflationary pressure by strengthening the yen, but that would hurt the economy.”

Ueda is receiving treatment for a liver condition and did not attend the meeting or cast a vote. He is expected to return for the July meeting. This week’s meeting was chaired by one of the BoJ’s deputy governors, Ryozo Himino.

The afternoon press conference will be led by the bank’s other deputy governor, Shinichi Uchida, whose comments will be closely scrutinised for signs of how the BoJ will continue to assess the negative economic impact of the Iran war.

>>> US After Hours Summary: HITI +18.2% sharply higher on earnings; PLAY -9.5% a

After Hours Summary: HITI +18.2% sharply higher on earnings; PLAY -9.5% and DOMO -1.8% lower on earnings; Dell +0.4% ticks higher on Air Force award

After Hours Gainers:

Companies trading higher in after hours in reaction to earnings/guidance: HITI +18.2%,

Companies trading higher in after hours in reaction to news: ARI +2.2% (update on strategic alternatives; determined that the dissolution of the company is advisable), IPI +2% (names new CFO), HDRN +1.5% (stock offering; also stock offering by holders), MDLZ +1% (appoints new CFO), TLN +0.9% (completes acquisitions of Lawrenceburg Power Plant, Waterford Energy Center and Darby Generating Station), NBIX +0.4% (presents first retrospective case series of CRENESSIT; also presents new VYKAT XR data at ENDO 2026), DELL +0.4% (awarded $1.4 bln Air Force call order for the Microsoft Enterprise License Agreement renewal), CRK +0.3% (sold a minority interest in Pinnacle Gas Services for $600 mln), GROY +0.3% (additional interest in an existing royalty over the REN project), HPE +0.2% (advances quantum computing at scale with expanded industry collaborations), AMAT +0.2% (introduces two new chipmaking systems), NEM +0.2% (appoints new CFO and COO), VSTM +0.1% (to report updated data and progress across VS-7375 Oral KRAS G12D (ON/OFF) Inhibitor TARGET-D Clinical Program), LMT +0.1% (awarded a $223.9 mln modification to Navy contract)

After Hours Losers:

Companies trading lower in after hours in reaction to earnings/guidance: PLAY -9.5%, DOMO -1.8% (also update on strategic alternatives)

Companies trading lower in after hours in reaction to news: ADPT -7.2% (convertible notes offering; also to pursue a separation of its Minimal Residual Disease and Immune Medicine businesses), NEO -5.4% (convertible notes offering), MAC -4% (stock offering), SFIX -3.5% (appoints new Chief Product and Technology Officer), LOVE -0.6% (appoints new CFO), QCOM -0.4% (in discussions to expand AI chip capabilities with potential Tenstorrent acquisition, according to The Information), SU -0.4% (files mixed securities shelf offering), GILD -0.3% (FDA accepts application for investigational once-weekly yeztugo), SSRM -0.1% (authorizes additional $500 mln for share repurchases)

(ZH) What Could Break The Bull Market This Summer

What Could Break The Bull Market This Summer

Key Takeaways
  • After nine straight up weeks, the bull market pullback we flagged finally arrived, and it stopped cold at the 50-day moving average.
  • The selloff reset an overbought tape without breaking trend. RSI fell from above 70 to the low 40s, and Thursday’s bounce came on broad participation.
  • Our Money Flow Breadth Ratio ticked up to 60%, back in buy territory, and we’re holding equity exposure at 100%.
  • The bigger risks haven’t gone anywhere: record margin debt, fading retail demand, and a 10-year Treasury that now out-yields the S&P 500.
  • This sets up more upside for now. It does not erase the odds of a deeper correction this summer if forward earnings expectations crack.
Two weeks ago, after the S&P 500 logged its ninth consecutive weekly gain, we discussed that a bull market pullback was coming. It came. From the May 27 record near 7,621, the index slid 4.5% and bottomed almost exactly on its 50-day moving average before ripping back to close Friday at 7,431.46. That is not the opening act of a bear market. That is the kind of bull market pullback that resets sentiment and, more often than not, clears the runway for the next leg higher. The harder question is what happens after the bounce.
The Correction We Told You To Expect
Make no mistake, I have been warning about the potential for a pullback over the last few weeks and repeatedly discussed taking profits and rebalancing risk. As I wrote in “Two-Month Market Rally: What Comes Next,” a market that climbs for 9 straight weeks gets stretched, and stretched markets tend to mean-revert. The only real questions were the “when” and “how much.” We suggested a bull-market pullback of 3% to 5%; toward the 50-day moving average, would be most likely. However, a larger correction is still possible. As noted, the actual decline ran 4.5% peak to trough and found its floor exactly where trend-followers add rather than abandon.
Notably, the dip buyers showed up on cue. When the S&P probed the mid-7,200s on Tuesday and Wednesday of last week, the same crowd that has bought every dip since the April 2025 low stepped in again, and by Friday the index had clawed back roughly a quarter of the prior week’s 2.64% drubbing. That close at 7,431.46 leaves the larger uptrend fully intact, sitting about 2.5% below the high rather than careening away from it.
Crucially, the setup still favors the bulls, at least for now.
First, the damage was technical, not structural. The 14-day RSI ran above 70 at the late-May high and fell to the low 40s at this month’s lows before settling back near 53. In plain terms, the market burned off its overbought condition without violating the trend, which is textbook.
Second, the quality of the bounce mattered. Thursday’s 1.75% surge came on broad participation rather than three megacaps doing all the lifting, and broad thrusts off support tend to mark real lows instead of dead-cat bounces.
Third, our own money-flow work agrees. The Money Flow Breadth Ratio (MFBR) is a rules-based model that “systematically adjusts portfolio equity exposure in response to the direction and persistence of institutional capital flows.” We use this analysis to size equity exposure in portfolios, and the MFBR ticked up to 60% as of June 12 and sits back in buy territory after sliding to 55% the prior week. The trailing four-week net flow has swung sharply positive following a deeply negative stretch, which historically reads as a contrarian buy. We’ve held exposure at 100% since April 17, and this signal keeps us there.
“As of June 12, 2026, with the S&P 500 at 7,431.46, the Money Flow Breadth Ratio (MFBR) stands at 60% and rising. This places the indicator in BUY territory (60-70%), triggering a NEUTRAL signal. The prior week reading was 55%, representing a 10% decline over the trailing four weeks. The model currently recommends HOLDING exposure at 100%, a level that has remained since April 17, 2026 (8 weeks). This reflects a FLOW-OVERLAY OVERRIDE: the trailing 4-week net dollar flow has swung sharply positive (>$300B) after a deeply negative prior 4 weeks, a historically strong contrarian buy signal.” – Bull Bear Report June 13th
The map from here is simple. Overhead, the 20-DMA at 7,466 is the first hurdle, then the round 7,500 mark, then the 7,621 record. Below, the 50-DMA at 7,248 is the line in the sand, with the 38.2% retracement at 7,118 and the rising 200-DMA at 6,882 beneath it. Hold 7,248 through Wednesday’s Fed meeting, and this stays a routine shakeout inside an uptrend.
What Could Break The Trade This Summer
None of that means you switch off your risk management. As we warned in “Leadership Is Narrow” and again in “Market Correction Risk,” the ingredients for a deeper drawdown are quietly building underneath a rising tape. Three of them deserve your attention.
Start with leverage. FINRA margin debt hit a record $1.30 trillion in April, up better than a third in a year, and now runs near 4% of GDP against a long-run median closer to 1.5%. Measured against M2, it’s back near the peaks that preceded the 2000 and 2007 tops. Borrowed money cuts both ways. It is an accelerant, not a cushion.
Next, watch the retail bid. Vanda’s flow data shows single-stock retail net turnover rolling over into negative territory in recent sessions, even as prices grind higher. That divergence matters. When the buyer who powered this rally starts selling into strength, the marginal source of demand thins out right as the supply picture gets heavier.
Then there’s the cold math on bonds versus stocks. The 10-year Treasury now yields about 4.45%, while the S&P 500’s trailing earnings yield sits near 3.7%. For the first time in this cycle, a risk-free Treasury pays you MORE than the index earns. That flips the equity risk premium negative and hands every allocator a credible, paid-to-wait reason to trim equities into bonds.
Layer on the supply story as detailed in “Equity Supply Surge”: Alphabet’s $80 billion secondary, SpaceX’s $75 billion IPO, and a queue of mega-raises from OpenAI, Anthropic, and the hyperscalers mean the market has to absorb a wall of new paper. More shares chasing the same dollars is a persistent headwind, not a one-day shock. As we noted in “Parabolic Semiconductor Rally,” when the most prized names all rush the exit at once, it pays to ask who is selling.
How We’re Positioning
So how do we square a buy signal with a real list of worries? We hold exposure and manage risk simultaneously. Those two things aren’t in conflict. The MFBR keeps us invested because the weight of the evidence and a clean test of support still point higher. Bob Farrell’s fourth rule reminds us that exponential moves tend to run further than anyone expects and then correct violently. Markets like that never correct gently. Therefore, we keep trailing stops disciplined, we refuse to chase the SpaceX-fueled enthusiasm at the highs, and we watch 7,248 like a hawk.
The calendar adds a second reason for that discipline. We’re walking into the weakest stretch of the year. As I detailed in “Market Correction Risk: Why Summer 2026 Looks Risky,” the May-through-October window has produced an average S&P 500 gain of just 1.7% since 1950, compared with better than 7% in November through April. The old “sell in May” line gets mocked every spring by people who haven’t looked at the data. The data is one-sided.
Then stack the election cycle on top. 2026 is a midterm year, and midterm years are the weakest and most volatile leg of the four-year presidential cycle. Jeff Hirsch’s Stock Trader’s Almanac has tracked the pattern for decades, and the numbers are sobering. Going back to the early 1960s, the average intra-year drawdown in a midterm year runs around 17% to 18%. That is well above the roughly 13% you see in the other three years. Notably, volatility tends to build up ahead of the November vote as investors handicap the balance of power in Congress.
Here’s the part that keeps me constructive, though. That midterm weakness has historically been a setup, not an ending. The 12 months after a midterm election have delivered an average S&P 500 gain of more than 12%. Furthermore, the Dow has climbed by more than 45% on average from its midterm-year low to its pre-election-year high. So the same seasonal soft patch that turns a routine bull market pullback into a deeper summer correction has, time and again, been the launchpad for the next leg up. We manage risk now, NOT because the bull market is over. We do it because we want dry powder and a steady hand when the seasonal low shows up.
None of this is about going to cash and hiding. It’s about tilting the book so you can sit through a noisy summer and still have ammunition for the fall. Here’s the playbook we’re running.
Howard Marks said it best.
“The riskiest thing in markets is the belief that there is no risk.”
With high-yield spreads pinned near 300 basis points, the market is pricing almost none. That’s exactly the backdrop where this kind of playbook earns its keep. Stay invested, but keep one hand on the exit.
The catalyst that turns a healthy pullback into something deeper won’t be a single oil-soaked CPI print. It’ll be the moment forward earnings expectations start to roll over while valuations sit at the high end of history. We aren’t there yet. Watch the Fed on Wednesday, watch wages, and watch whether second-half earnings estimates hold. The trend is your friend right up until the day it isn’t. Our job between now and then is to stay invested without going blind.
What’s your read? Are you adding to this dip or trimming into strength? Does the gap between a bullish tape and a long list of risks have you second-guessing your own positioning? If so, that’s exactly the conversation worth having. Connect with our team, and let’s pressure-test your portfolio before the summer does it for you.

(ZeroHedge) Could Trump's Fable 5 Export Curbs Slow China's AI Model Race

Could Trump's Fable 5 Export Curbs Slow China's AI Model Race

The biggest AI story to start the week is the Trump administration's decision to place export controls on Anthropic's Fable 5 and Mythos 5 models, forcing Dario Amodei's frontier AI lab to restrict foreign access.

The move comes as Chinese open-source models have been rapidly closing the compute gap with U.S. labs. Anthropic's latest release appears to have widened that gap again, particularly in frontier reasoning, coding, and cybersecurity use cases.

Export controls could have second-order effects on the global AI race, according to analysts at Jefferies. By limiting access to Anthropic's most advanced models, Trump officials may slow the pace at which foreign developers, particularly in China, can study, benchmark, or distill these advanced frontier models into cheaper open-source systems.

"US models are improving at a faster pace, likely due to computational advantage, but anti- distillation and US export control are new negatives for China AI," the analysts wrote in a note on Sunday.

The key question now is whether Fable 5 and Mythos 5 include stronger anti-distillation safeguards to prevent Chinese labs from replicating or compressing Anthropic's advances into open-source models. If so, the export curbs may not just be about access. They may represent a broader effort to protect America's AI lead.

To understand the full AI model landscape, not just in the West but also in the East, Bank of America analysts, led by Alex Liu, penned an insightful note on Monday morning about leading Chinese models.

Liu wrote that China's AI model market is moving into a two-speed global structure, with U.S. labs likely to retain the lead in frontier capabilities while Chinese players gain share in lower-cost, high-volume use cases.

She noted that Chinese models are narrowing the gap through efficiency gains, architecture optimization, distillation, and lower-cost inference, making them increasingly affordable.
Liu said AI labs at Alibaba, ByteDance, Tencent, and Baidu are racing against independent labs, such as DeepSeek, Zhipu, MiniMax, and Moonshot AI.
Who's who in the China AI model market
Incumbents
  • Major internet companies such as ByteDance, Baidu, Alibaba, and Tencent—many of which have established cloud businesses—have developed proprietary AI foundation models in-house.
Independent AI labs
  • DeepSeek, MiniMax, Zhipu, and Moonshot AI are independent AI labs
China AI model landscape is intensely competitive
Investing landscape of Chinese AI labs
Liu's view is that China AI has shifted from a frontier story to an affordability story. But with the U.S. government now able to halt foreign access to advanced models, as it just did with Anthropic's Fable 5, it appears increasingly difficult for Chinese labs to copy, distill, or reverse-engineer U.S. frontier models

TechCrunch : A satellite just learned to find things on its own — here’s what th

A satellite just learned to find things on its own — here’s what that means

For the first time, an Earth observation satellite has found what it was looking for — on its own, without human analysts on the ground. The milestone, which occurred in April, marks the first reported use of a vision-language model in orbit, and offers a glimpse of how AI could fundamentally change what space-based sensors are capable of — and how much they’re worth.

Typically, satellites download large chunks of data to analysts on the Earth below, who use machine learning algorithms or their own eyes to figure out what’s going on. But onboard YAM-9, a spacecraft built by space infrastructure company Loft Orbital, a software package built by NASA’s Jet Propulsion Laboratory identified areas of interest in response to natural language queries.

Google DeepMind’s Gemma 3 — the vision-language model, or VLM, that powered the demonstration — is purpose-built for edge applications, meaning it is designed to run on limited hardware far from a data center. VLMs combine the contextual understanding of large language models with the ability to analyze imagery: Researchers asked the model to classify sensor data where natural environment meets human development, for example, or to identify infrastructure around railway hubs — and it did.

The demonstration is significant for two reasons. In the near term, it could make space sensors far more useful by doing initial data triage on orbit, reducing the flood of raw data that analysts currently have to wade through. Longer term, it’s a proof point toward running larger-scale AI infrastructure in space.

“It opens the door to always-on, patrol layers in space,” Loft’s head of AI, Paul Lasserre, told TechCrunch. “If you have a VLM, you can have logic — like ‘monitor this border for me, and let me know when something is suspicious,’ and interact back and forth with the satellites.”

Loft’s spacecraft are designed as platforms for third-party customers. The business model is closer to infrastructure-as-a-service than traditional satellite manufacturing. One recent deal saw it build, launch, and operate six new satellites for EarthDaily, which will analyze and market the data collected onboard the spacecraft. YAM-9 was launched in the fall of 2025 as a pathfinder for the company’s orbital AI projects, and includes a Nvidia Jetson Orin AGX GPU, one of the leading chips used in space compute.

Juan Delfa Victoria, a technical leader in NASA JPL’s AI group, led the development of NAVI-Orbital, a software package that was effectively the harness for the Gemma 3 VLM. While Gemma 3 is off the shelf, software engineers had to streamline the software package to reduce the amount of libraries and memory it would require.

While this is the first reported use of a VLM on orbit, we can expect other companies to follow suit. Planet Labs flies satellites with Jetson Orin processors; for now, it is using them for simpler object detection tasks, but a spokesperson says research is underway on other AI applications, including VLMs.

Kepler Communications, which operates the largest group of GPUs in space, declined to say whether it had deployed VLMs in space due to NDA agreements with partners, but noted that there have been “several undisclosed use cases of our compute environment” since those spacecraft launched in January.

“Now that we’ve proven the concept, that’s really the direction of travel,” Lasserre said. The goal is to build out the constellation to ensure real-time coverage of anywhere on Earth, which which he says would take somewhere between 50 and 100 satellites like YAM-9. (Loft currently operates 12 spacecraft on orbit.)

Lessons learned deploying these smaller models on orbit will inform how companies attempt to deploy larger-scale compute infrastructure in space, particularly in the prosaic-but-vital areas of power and memory management.

They could also pave the way for new scientific tools. The idea for NAVI-Space began when Delfa Victoria and JPL researcher Taran Cyriac John were thinking about digital assistants for astronauts exploring the moon or Mars.

“We’re thinking, okay, you have astronauts with pressurized suits, and you know they cannot be tapping on a keyboard, whatever they want to do is complex.” Delfa Victoria said. “So, how about we provide an assistant, like in video games and in movies, where you see an AI which is interactive?”

Just don’t call it HAL 9000.

The Information : Inside Broadcom’s Bold Move to Boost Demand for Its Chips

Inside Broadcom’s Bold Move to Boost Demand for Its Chips

The Takeaway
  • Broadcom backs $35 billion AI chip order to boost demand.
  • Broadcom’s financial backstop for chip lease impacts credit.
  • Company plans to finance over 20 GW of compute capacity.

When Broadcom last week announced a funding venture with Apollo and Blackstone to pay for a gigawatt of computing capacity to be used by Anthropic, it looked like the latest in a series of AI computing deals funded by private equity. Behind the scenes, however, the deal represents a risky move by Broadcom to boost demand for its chips.

While the announcement didn’t spell out Broadcom’s role, the company—which works with tech firms such as Google to design their AI chips—is providing a key financial backstop for the $35 billion order of chips. In doing so, it is following in the footsteps of Nvidia, which has employed similar vendor financing techniques to accelerate its own chip sales. But Broadcom doesn’t have Nvidia’s deep pockets, making it a bigger gamble.

Indeed, as recently as March, Broadcom CEO Hock Tan was reluctant to use Broadcom’s balance sheet to provide such a backstop, according to a person who spoke with Broadcom executives. But he changed his mind. Broadcom approached Morgan Stanley for help finding a way to finance chip sales while limiting the debt it must add to its balance sheet, said a person involved in the transaction.

Tan had reason to be cautious. Broadcom generates a fraction of the cash produced by Nvidia, which has provided similar backstops. Broadcom also had $65 billion of debt and just $19.6 billion of cash as of May 3, it reported earlier this month.

But if Broadcom didn’t take this step, it was at risk of getting left behind in the AI chip race.

“We are at a historic inflection point where the demand for AI compute is fundamentally reshaping the global economic landscape,” Tan said in this week’s announcement, hinting at the context behind his apparent reversal. He went on to call it a “once-in-a-lifetime opportunity.”

Even so, one institutional investor who owns Broadcom bonds raised their eyebrows this week when asked about the deal, noting that the “big question” they had was how the company planned to account for these types of guarantees on its balance sheet. Though not traditional debt, the obligations still count as liabilities that must be reported on the balance sheet.

The transaction will have a “modestly negative impact” on Broadcom’s credit due to the “debt-like obligation” of the backstop, S&P Global Ratings wrote in a note.

To be sure, Broadcom’s AI chip business has boomed in recent quarters, lifting its profits. The company reported in early June that its revenue for the second fiscal quarter grew 48% as AI chip revenue jumped 143%. Free cash flow rose 60%.

Broadcom didn’t respond to requests for comment.

A Larger Plan

More deals are likely. Broadcom described this one as the first transaction in a larger plan to work with Apollo and Blackstone to finance more than 20 GW of compute. That suggests its ambitions, if fulfilled, could see it providing future guarantees on deals worth a combined $700 billion in chip purchases, although the company’s backstop is very limited.

With so much to come, Broadcom’s initial transaction may provide a model for future structures. This account is based on conversations with four people with knowledge of the transaction, who asked for anonymity to discuss details beyond what the press releases included.

At its most basic level, the people said, the transaction involves a special purpose vehicle that will own the chips—tensor processing units Broadcom co-designed with Google–and lease them to Anthropic. To get the money to buy the chips, the SPV sold $35 billion in bonds to a collection of lenders led by Apollo and Blackstone.

Both investment firms kept some of the bonds for their own funds—or in Apollo’s case its captive insurance arm—and then sold the rest to outside investors, two of the people said. Apollo bought more than half, while Blackstone accounted for the rest, one of them said.

All of the notes—sold in three tranches of varying sizes—are secured by the chips. If Anthropic can’t pay the lease, the lenders have the right to sell the chips to recoup their money.

However, there’s still uncertainty around how quickly AI chips will decline in value over their usable lifetimes due to factors like new chips making old ones obsolete, or demand for AI declining in the future. If the residual value of the chips at the end of the lease falls below a certain level, Broadcom has agreed to compensate lenders holding the two safer tranches.

Broadcom’s backstop ensured that those two pieces secured an investment grade rating and yield around 5.75%, one of the people said.

The third tranche does not benefit from Broadcom’s backstop. That means if Anthropic defaults on the lease and the sale of chips doesn’t fully repay the debt, the investors who own those notes would be exposed to Anthropic’s credit risk. Those notes are rated below investment grade and yield 8.5%, one of the people said.

Dual Contingency

The structure means Broadcom is protected by a dual contingency. For its backstop to kick in, Anthropic has to default on the lease and the chips have to be worth less than what investors were guaranteed.

The notes are fully amortizing over the five-year term, meaning Broadcom’s risk declines to zero over the term, one of the people said.

“The strategic rationale for Broadcom has to do with supporting the end demand from customers like Anthropic for its chips—using its scale and dry powder, from a credit perspective, to put itself in a prime proposition such that Anthropic is using its products,” one of the people said. “That concept is not new or uncommon; many manufacturers across many industries have captive finance companies that help their customers purchase and finance their goods.”

This is one of the first times bond investors have been able to make a bet on Anthropic, which isn’t profitable and hasn’t issued corporate debt, one of the people said.

Banks, insurance companies, credit investors and others bought into the safer portions of the deal, assigned A1 and A2 monikers, while opportunistic credit investors bought the riskier B notes, one of the people said.

On the A1 tranche, Wells Fargo served as the global coordinator, while BNP Paribas, Citi and UBS served as joint book runners and lead arrangers, according to an Apollo press release. Goldman Sachs, Bank of America and Morgan Stanley served as joint placement agents on the A2 tranche.

In this structure, Atlas SP Partners, a unit majority owned by Apollo, formed the SPV and consolidated it on its balance sheet.

For Anthropic, the arrangement represents the clearest sign yet that the company is forging ahead with plans to create its own supply of compute, rather than renting facilities and chips from other cloud providers like Google or Amazon.

Chip financing deals like this are contingent on the company having space in data centers where it can install the chips once it has purchased them. To that end, Anthropic has lined up Google to provide a backstop on the AI lab’s leases for five data center facilities, The Information reported.

“Whether it is a compute contract or a chip lease,” one of the people said, “the lease start date is dependent on the data center being ready.”

FT : Chipmaker Nvidia seeks to raise over $25bn in first bond deal since 2021

Chipmaker Nvidia seeks to raise over $25bn in first bond deal since 2021
Debt sale set to test investor appetite for further exposure to AI sector amid a deluge of borrowing

Chipmaker Nvidia is planning to sell $25bn of investment-grade debt in the US on Monday, its first bond sale in five years, in a test of investor appetite for further exposure to the AI sector.

In a marquee seven-part bond offering, the company will issue a wide range of maturities from two years to 30 years, according to a term sheet seen by the FT.

The issuance was upsized from $20bn after receiving more than $85bn in orders by early afternoon in New York, according to people familiar with the deal. 

Thanks to robust demand, the 10-year portion of the bond was expected to yield 0.5 percentage points above US Treasuries, down from 0.75 percentage points during initial discussions, one of the people said.

Favourable market conditions after the US-Iran deal are allowing Nvidia to raise debt at a relatively low cost, said Lauren Wagandt, a portfolio manager at T Rowe Price. 

“It’s a very high-quality company at the end of the day,” said Wagandt. “And it doesn’t come to the market as often as the other tech names.”

The issuance by the semiconductor group, the biggest beneficiary of Big Tech’s trillion-dollar spending spree on AI infrastructure, comes as tech groups race to secure funding amid an intensifying AI arms race, but also as Wall Street faces a torrent of new equity and debt issuance, including SpaceX’s record $75bn initial public offering.

“We intend to use the net proceeds from this offering for general corporate purposes, including repayment and refinancing of outstanding notes,” Nvidia said.

Monday’s offering is at least three times larger than Nvidia’s previous bond sale in 2021 during the coronavirus pandemic, when it raised about $5bn. When completed, it will more than triple Nvidia’s debt outstanding to about $30bn from the current level of $8.5bn. 

Early signs of market fatigue have prompted some tech companies to find alternative avenues for financing.

Anthropic has turned to private credit investors to seal a $35bn deal backed by Broadcom. Google’s parent Alphabet decided to issue equity for the first time in more than two decades, bringing in $85bn in fresh capital earlier this month.

Nvidia’s position as the AI industry’s go-to supplier of the powerful chips needed to build large language models such as OpenAI’s GPT has proven extremely lucrative for the Silicon Valley company, with its free cash flow in the year to January leaping 59 per cent to $96.6bn.

However, after its valuation peaked at about $5.7tn in May, its shares have fallen alongside the wider semiconductor market in recent weeks, with its market capitalisation dropping below $5tn at the end of last week.

While reaping huge profits from AI spending, Nvidia has also become a significant investor in AI companies, committing a total of more than $90bn to developers including OpenAI, Anthropic and xAI, and suppliers including Coherent, Marvell, Lumentum and Corning. In some cases, it has also agreed to act as a backstop or financial guarantor to customers building cloud computing services using its chips, including CoreWeave and Nscale.

The increasing use of financial guarantees and the interdependence of AI companies has raised concerns about concentrated risks among bond investors, said Tom Murphy, global head of investment-grade credit at Columbia Threadneedle Investments. 

“The market has started to get worried about these circular financings, because if somebody in that ecosystem is having a problem, then the whole thing could be a problem,” Murphy said.

Nvidia has a double-A credit rating, the third-highest score. More indebted AI player Oracle sits just two notches above a junk rating. 

Goldman Sachs, JPMorgan and Morgan Stanley are active bookrunners of the transaction.

WWD : Jacquemus Confirms Show Venue in Corsica

Jacquemus Confirms Show Venue in Corsica
The designer has selected a remote rocky outcrop with a lighthouse and staggering sea views.


SEEING THE LIGHT: Returning to a gobsmacking natural setting for his next fashion show, Simon Porte Jacquemus has selected a rocky island off Corsica whose main landmark is a small, white lighthouse.

The coed collection for spring 2027, titled “Le Bonheur,” is to be unveiled at 10 a.m. on June 29.

“I am deeply honored and proud to present this show at the Phare de la Pietra in Île-Rousse, a place of exceptional beauty and history,” Jacquemus said in a statement shared exclusively with WWD.

“I am also delighted to partner with the town of Île-Rousse in supporting the restoration and preservation of this unique heritage site,” he added. “Bringing these places to life and helping ensure their legacy is passed on to future generations is something that is particularly close to my heart.”

A happy few were sent a save-the-date for the show, which comes just after the end of men’s fashion week in Paris.

Jacquemus said the guest list would be “very intimate and will only include members of the press, friends of the house, a few buyers as well as international and local celebrities.”

The house also noted it is organizing a casting in Corsica to select local models for the runway display.

Famed for his transporting destination shows — in lavender fields, salt flats or grand French gardens — Jacquemus has recently favored indoor locations, including the Picasso Museum in Paris and the private apartment of architect Auguste Perret.

Built in 1857 during the reign of Napoleon III, the lighthouse serves as the entrance to the northwest coast.

WWD : Philippe Stern, Patek Philippe’s Honorary President, Dies at 88

Philippe Stern, Patek Philippe’s Honorary President, Dies at 88
He guided the company through the quartz crisis of the 1970s and was a staunch believer in high-end mechanical watchmaking.

Philippe Stern, the watchmaking executive who guided Patek Philippe through the quartz crisis and was the third generation at the helm of the family-owned company, died Sunday at age 88.

The company revealed his death on Monday.

Born in 1938, Stern was a grandson of Charles Stern, who acquired Patek Philippe with his brother Jean Stern in 1932 after being long-term dial suppliers of the watchmaker, and the son of Henri Stern, who developed the company’s international distribution, and particularly the U.S. market.

He spent his entire career at the family-owned firm, first in New York where he joined, after studying economics and commerce, its Henri Stern Watch Agency subsidiary in 1963.

Returning to Geneva in 1966, he spent the next decade working his way through the company, including overseeing the launch of the Nautilus sports watch, which remains a reference in watchmaking.

Stern became the company’s managing director in 1977, as the Swiss watchmaking industry faced the quartz crisis. A staunch believer in mechanical watchmaking, he embraced craftsmanship and pursued ambitious projects such as the nine-year-long development of the Calibre 89 with 33 complications.

In 1993, Stern became president of the company. Under his leadership, the company pursued a strategy of vertical integration, culminating with the state-of-the-art Plan-les-Ouates manufacturing site in 1996.

After passing the baton to his son Thierry Stern in 2009, he remained as honorary president and member of its management committee.

A keen collector, Stern also amassed a rich collection that spanned examples of the brand’s timepieces but also pieces that retrace the history of watchmaking since the 16th century. They became the basis for the Patek Philippe Museum in Geneva, which he inaugurated in 2001.