Broadcom Climbs to $367 on $100B AI Financing Report as Stock Sits 21% Below Its June Peak

Broadcom traded $367.06 in Friday's premarket, up 0.83%, and held $366.92 into the session with a range of $360.55 to $368.00 on volume of 19.73 million shares against a 19.77 million average. The stock closed Thursday at $362.48 and rose as much as 1.1% in late trading after Bloomberg reported the company is negotiating a debt package that could reach $100 billion.

The move is small. The context is not.

At $367.06 the stock sits 25.8% below its 52-week high of $495.00 and 28.2% above the low of $286.17. Market capitalization stands at $1.73 trillion with a trailing P/E of 60.34 and a dividend yield of 0.70%.

The drawdown has been methodical. Broadcom has fallen roughly 21% from its early-June peak, with about 12% of that coming over seven trading sessions and more than 7% across the last five. The slide began after the fiscal second-quarter report on June 3 and has not stopped since.

The technical picture heading into Friday was broken on the short timeframes and marginal on the long one. At $364.17 the stock was trading 8% below its 20-day simple moving average of $398.16 and 5.8% below its 50-day at $389.14. It was only 0.8% below the 200-day at $369.20 — the line that separates a correction from a trend change.

Friday's premarket move puts the stock back through that 200-day. Whether it holds is the entire near-term question.

The disconnect between the operating story and the price is what makes this name interesting right now. AI semiconductor revenue hit $10.8 billion in the fiscal second quarter, more than doubling year-over-year, with management guiding to $16 billion in the current quarter. The company has line of sight to more than $56 billion in AI revenue this fiscal year and over $100 billion next.

The stock is down 21% anyway.

Published consensus targets sit far above the current price — a Buy rating with an average forecast of $513.68 in one survey and $493.24 in another, against 42 analysts averaging a figure 24% above the recent close of $360.

That gap is either an opportunity or a stale model. The September 2 report settles it.

The Bloomberg Report: $60–70 Billion Senior, $30 Billion Junior, One SPV

The headline that moved the stock Thursday evening is the largest single corporate financing discussion in the AI buildout so far.

Broadcom is in talks with a group of lenders to raise more than $60 billion in debt for an AI chip financing deal that would benefit Anthropic and other companies. The structure under discussion includes a senior-secured tranche of roughly $60 billion to $70 billion plus a junior tranche of approximately $30 billion, taking the total as high as $100 billion.

The debt would be issued through a special-purpose vehicle. Broadcom would guarantee part of the senior tranche.

Blackstone and Apollo Global Management are in discussions to participate, building on a partnership the three companies struck in June. Talks remain ongoing, terms may change, and the financing could roll out in stages rather than all at once. Representatives for Broadcom, Anthropic, Apollo and Blackstone declined to comment.

The scale needs context to be understood. Broadcom held $19.63 billion in cash and equivalents as of May 3, 2026, against $62.66 billion in long-term debt. A $100 billion financing is 1.6 times the company's existing long-term debt load and roughly 5.8% of its $1.73 trillion market capitalization.

Critically, the SPV structure means most of it does not appear on Broadcom's balance sheet. That is the model Meta established with its off-balance-sheet vehicle earlier in this cycle, and it has become the template for AI infrastructure finance.

The company is not borrowing to build. It is borrowing so its customers can buy.

The purpose is explicit: the financing would help firms including Anthropic secure Broadcom chips and other AI infrastructure. Broadcom sells more silicon and networking equipment; the customer gets capacity without funding it from operating cash flow; the private credit market absorbs the paper.

This is the same architecture Nvidia used earlier this month when a coalition including BlackRock and Goldman Sachs lined up financing for compute purchases. The two dominant AI chip suppliers are now both financing their own demand.

That parallel is why the deal moved Broadcom's stock less than the number suggests it should have.

AI XPV: How The First $35 Billion Deal Actually Worked

The proposed financing follows a template that already exists and is already trading.

On June 9, 2026, Broadcom established the AI XPV Platform with Apollo and Blackstone's Credit and Insurance business as initial anchor investors. The platform is designed to enable more than 20 gigawatts of compute capacity using Broadcom XPUs and networking solutions customized for frontier AI labs including Anthropic and OpenAI, through 2028.

It launched with an initial tranche of approximately $35 billion, facilitating Anthropic's previously announced expansion of more than one gigawatt of compute infrastructure for training and inference, deploying at Fluidstack-based sites from mid-2026.

The mechanics are the important part. The special-purpose vehicle finances the purchase of custom AI chips developed by Google and Broadcom, then leases the hardware to Anthropic. Broadcom provides a backstop for Anthropic's payment obligations on the largest senior tranches.

That backstop is what made the economics work. Broadcom's guarantee enabled the senior debt tranches to win investment-grade ratings, which kept borrowing costs low. Without it, financing an unrated private company's chip purchases would have priced at high-yield levels.

The transaction uses a delayed-draw format, allowing the borrower to access money as needed. There are approximately 16 separate releases scheduled over a little more than a year as chips are produced. The financing was initially issued through the private placement market.

By early 2027, roughly $15 billion of the debt is expected to be accessible to investors in the secondary market — which means this paper will start trading and generating a visible price signal on how the market values Broadcom's guarantee.

The scale of what remains is the sobering part. The partnership plans to finance more than 20 gigawatts of computing power, which would require hundreds of billions of dollars and would roughly equal the output of 20 nuclear plants. The first $35 billion commitment adds one gigawatt.

On those figures, the partnership has funded about a twentieth of its stated target.

A $100 billion follow-on would take the cumulative total to $135 billion and, at the same conversion ratio, roughly four gigawatts. That is still a fifth of the 2028 goal.

Broadcom Is Now Financing Its Own Demand

The structural question every shareholder should be asking is whether this is a growth engine or a growing obligation.

The bull framing is straightforward. Broadcom's custom AI chips — application-specific integrated circuits — have become the preferred alternative to general-purpose GPUs for hyperscalers running large, predictable workloads. That positioning has produced real numbers: AI semiconductor revenue reached $10.8 billion in the fiscal second quarter, an annualized run rate near $43 billion, with management guiding to $16 billion in the current quarter.

Financing removes the single largest friction in converting that demand into orders. An AI lab with a compute roadmap but a finite balance sheet cannot buy four gigawatts of silicon from operating cash flow. Vendor-arranged financing lets the order land now and the payment stretch across the useful life of the hardware.

By easing customer financial burdens, Broadcom secures orders it would otherwise wait years for.

The bear framing is equally clear. Broadcom is guaranteeing payment obligations for companies whose revenue models are still being established. If an AI lab's compute demand falls short of its roadmap, or its funding environment tightens, the guarantee converts from a marketing tool into a liability.

The counterparty quality has improved substantially. Anthropic has been reported to be targeting an IPO at SpaceX scale, citing a $65 billion revenue run rate and a first profitable quarter, and its most recent private round valued it at $350 billion after raising more than $10 billion — more than double its initial target. That is not a startup credit.

The concentration is the harder problem. A small handful of large AI customers drive the bulk of Broadcom's revenue and future growth. Guaranteeing their debt on top of selling them chips doubles the exposure to the same names.

The Apple agreement provides some diversification, expected to exceed $30 billion and run through 2031. OpenAI has separate custom chip agreements.

But the shape of the risk is unchanged: Broadcom's earnings, its order book and now its contingent liabilities all point at the same four or five counterparties.

That is why the equity has not rerated on the financing news the way it might have a year ago.

The CDS Widened And That Is The Signal Worth Watching

There was a reaction to Thursday's report that had nothing to do with the equity.

Broadcom's credit default swaps widened as the market absorbed the scale of the off-balance-sheet borrowing under discussion. That is the credit market pricing the guarantee rather than the growth.

The distinction matters. Equity holders see a $100 billion financing as demand enablement — more chips sold, more networking gear shipped, more revenue converted from pipeline to backlog. Credit holders see a company with $62.66 billion of existing long-term debt and $19.63 billion of cash taking on contingent obligations that could exceed its entire balance sheet.

The SPV structure is designed to keep those obligations off the reported balance sheet. Rating agencies and credit analysts do not treat them that way. A guarantee is a guarantee regardless of where the vehicle sits.

The AI XPV precedent shows exactly how much the guarantee is worth. It converted debt that would otherwise have priced as high-yield into investment-grade paper. That spread compression is the entire economic value Broadcom contributes to the structure — and it is a direct transfer of Broadcom's own credit quality to the borrower.

Do it once at $35 billion and it is a marketing expense. Do it repeatedly toward $135 billion and it becomes a leverage decision.

The unprecedented scale of borrowing now under discussion underscores the capital requirements of the AI boom, which has prompted a slew of novel debt deals at a pace and scale that has simultaneously unnerved some market participants.

That unease is the counterweight to the equity story. If AI infrastructure credit spreads widen materially, the financing model that converts Broadcom's pipeline into revenue stops working, and the order book that supports a $100 billion FY27 AI target becomes conditional rather than committed.

For now the paper is placing. Roughly $15 billion of the first deal reaches the secondary market by early 2027, and that will be the first public price on how the market values a Broadcom guarantee.

Watch that spread more closely than the stock.

The June Quarter: $22.19 Billion, A Revenue Miss, And A 15% Drawdown

Everything that has happened to this stock since June traces to one report.

Broadcom posted fiscal second-quarter revenue of $22.19 billion against $22.27 billion expected — a miss, despite 48% year-over-year growth from $15 billion. Adjusted EPS came in at $2.44 against $2.40 estimated, a 1.82% surprise. Net income reached $9.31 billion, up from $7.35 billion in the prior quarter.

AI revenue more than doubled year-over-year to $10.8 billion, attributed to custom AI chips plus the networking components needed to wire them together. Management guided AI revenue to triple in the current quarter to $16 billion.

The company authorized a $10 billion share repurchase and raised its dividend for the fifteenth consecutive year.

The stock fell more than 12% on the print and closed down about 15% the following session.

The cause was not the miss. It was what the CEO did not say. Hock Tan declined to raise the full-year target of $100 billion in AI chip sales, telling analysts the company expected momentum to continue into fiscal 2027 and reiterating — rather than lifting — the guidance for AI semiconductor revenue in excess of $100 billion.

There was a second disclosure that received less attention and may matter more. Tan said the company would offer chips only, instead of the complete integrated AI systems Broadcom had previously said it would provide to customers. That is a narrowing of the addressable content per rack, and it came without explanation.

The comparison against the prior quarter shows how high the bar had become. Fiscal first-quarter AI revenue was $8.4 billion, up 106% year-over-year and above the company's own forecast. Two consecutive quarters of triple-digit AI growth had trained the market to expect a guidance raise every time.

Six core custom chip customers drive the AI business, including Anthropic, Google, Meta and OpenAI.

The stock closed at $420.56 shortly before the drawdown began in earnest. At $367.06 it has given back 12.7% from that level and roughly 21% from the early-June peak.

Two months of selling on a quarter where AI revenue doubled.

$56 Billion This Year, $100 Billion Next — The Number Tan Won't Raise

The forward AI targets are the most important numbers in this story and they have not moved.

Management has outlined projections for AI semiconductor revenue of more than $56 billion for fiscal 2026 and over $100 billion for fiscal 2027. Those figures would substantially increase the share of Broadcom's business tied to AI XPUs and networking.

Work the arithmetic. First-quarter AI revenue was $8.4 billion, second quarter $10.8 billion, third quarter guided to $16 billion. That totals $35.2 billion through three quarters. Reaching $56 billion requires $20.8 billion in the fiscal fourth quarter — a 30% sequential increase on top of the 48% sequential jump already guided for Q3.

Then fiscal 2027 requires more than $100 billion against roughly $56 billion this year. That is 79% growth off a base that itself nearly tripled.

Those are extraordinary numbers and management has reiterated rather than raised them for two consecutive quarters.

Two readings are available. Either Tan is being conservative in the way he has been throughout his tenure, and the actual figures land above the guide — or the targets were set aggressively enough at the outset that there is no room to lift them, and the risk is to the downside.

The market has been trading the second interpretation since June 3.

The pipeline supporting the targets is real. Broadcom helped Alphabet build its Tensor Processing Unit, and in April the two signed a five-year deal covering future TPU generations plus the networking gear to wire them into data centres. The Apple agreement exceeds $30 billion through 2031. OpenAI has custom chip agreements. New client orders are expected to come online next year, extending Broadcom beyond the small handful of hyperscalers it currently serves.

Multi-year AI infrastructure supply commitments from Anthropic, OpenAI and Meta give clearer visibility on demand while heightening exposure to any change in those customers' spending plans.

Hyperscaler AI capital expenditure is forecast to jump 57% in 2027, which is the macro assumption underneath the entire $100 billion target.

If that capex number holds, the guide is achievable. If it slips, the guide is the ceiling.

Marvell Just Walked Into Google's TPU

The competitive development this week is the one the market has been most worried about, and it landed on August 19.

Marvell announced a TPU deal with Google. Broadcom shares dropped 2.3% on the news. Marvell surged after an agreement allowing Alphabet to buy up to $12.2 billion in Marvell shares.

That is a direct incursion into Broadcom's largest and most established custom silicon relationship. Alphabet is the anchor customer — the company Broadcom helped build the Tensor Processing Unit with, and the account that validated the entire custom accelerator business model.

An equity stake of up to $12.2 billion is not a procurement decision. It is Alphabet aligning a second supplier's incentives with its own roadmap, which is what a customer does when it wants to reduce dependence on a single vendor.

The strategic logic from Google's side is obvious. Broadcom's custom accelerator business is dominant enough that pricing power has shifted toward the supplier. Bringing in a credible second source resets that negotiation regardless of whether Marvell ever ships at scale.

The counterweight is that Broadcom and Alphabet signed a five-year agreement in April covering future TPU generations plus networking. That contract predates the Marvell deal and presumably contains volume commitments.

The sell-side framing this week has been to buy one and hold the other, with Broadcom on the hold side of that pairing at least one firm.

The broader semiconductor tape has been unhelpful. Friday saw Teradyne down 3.60%, FormFactor down 3.00%, Onto Innovation down 2.69%, Marvell itself down 2.55%, and Cerebras down 2.96%. Nvidia traded $214.86, down 0.92%, on a $5.204 trillion market capitalization.

Every one of those names carries a large twelve-month gain and every one is being sold as the 30-year Treasury holds 5.25% and long-duration multiples compress.

Ark Invest bought Broadcom and Cerebras on August 19 while trimming AMD — one of the few visible institutional buys during the drawdown.

Six Customers Carry The Entire Thesis

The concentration risk is not a footnote. It is the defining characteristic of this business.

Broadcom has six core custom chip customers driving its AI revenue growth, including Anthropic, Google, Meta and OpenAI. A small handful of large AI customers generate the bulk of revenue and essentially all of the future growth in the semiconductor segment.

That structure produces spectacular economics when the customers are spending and catastrophic operating leverage when they are not.

The customer set has been reshuffling. Anthropic has passed OpenAI as the hottest AI upstart by one measure, is reported to be targeting an IPO at SpaceX scale on a $65 billion revenue run rate with a first profitable quarter delivered, and its financing needs are what the $100 billion package is built around. OpenAI's most recent results were characterized as tepid, failing to match Anthropic's growth pace.

That divergence matters for Broadcom because the company's forward revenue is a weighted average of its customers' capital plans. If one accelerates and another decelerates, the aggregate can hold. If the two largest decelerate together, there is nothing underneath.

Alphabet is the largest and most durable of the group, with the TPU relationship dating back years — which is precisely why the Marvell development landed as hard as it did.

Meta contributes through multi-year supply commitments. Apple's agreement exceeding $30 billion through 2031 sits in a different category, tied to consumer silicon rather than data centre accelerators.

The structural offset is the software business. Broadcom is not a pure-play semiconductor company. Roughly 58% of revenue came from semiconductor products and 42% from infrastructure software as of last year, with VMware providing recurring, high-margin revenue that does not depend on AI capex cycles.

That mix is why the company has been described as a prolific generator of cash flow and a terrific aggregator of firms, with execution and operating efficiency that let it outperform its end markets organically.

The software leg is what makes a 60x trailing multiple defensible during an AI drawdown. It is also the leg the market has been ignoring for two months.

September 2: $3.16 On $29.44 Billion

The catalyst that resolves this is twelve days away.

Broadcom reports fiscal third-quarter results on September 2, covering the quarter ended August 2, 2026. Consensus expects earnings of $3.16 per share against $1.69 a year earlier — 87% growth — on revenue of $29.44 billion against $15.95 billion in the year-ago period. Some estimates put EPS at $3.21 and revenue at $29.23 billion.

The company's own guidance called for consolidated revenue growth of 84% year-over-year to $29.4 billion.

The number inside the number is AI semiconductor revenue guided to $16 billion, up from $10.8 billion in the prior quarter — a 48% sequential increase. That single line is 54% of total guided revenue, and it is what the stock trades on.

The setup is unusual because the stock has already fallen 21% into the print. Expectations have been reset lower by two months of selling, which changes the asymmetry. A quarter that merely meets guidance and reiterates the $100 billion FY27 target would be received differently at $367 than it would have been at $495.

What the market wants is a raise. The company has declined to lift the AI target for two consecutive quarters, and a third reiteration would confirm the ceiling interpretation.

The concerns going in are documented: high expectations, elevated Treasury yields compressing long-duration valuations, and a potential letdown around one large customer's financial progress.

The valuation frames the risk. At $367.06 against consensus FY26 earnings, the trailing multiple of 60.34 compresses substantially on forward numbers. The last quarter delivered $2.44 and this one is guided to $3.16 — annualizing the guided quarter puts the forward multiple near 29 times.

That is not an expensive multiple for a business growing AI revenue at triple-digit rates with a 42% software mix underneath.

VMware Explore runs August 31 to September 3 at The Venetian, overlapping the earnings date. Product announcements from the software side could add a second narrative that has been absent all summer.

Twelve days, one number, and a stock that has already priced disappointment.

Technicals: The 200-Day At $369.20 Is The Line

The chart has been deteriorating for two months and it is now sitting on the level that decides the trend.

At $364.17 on Thursday, Broadcom traded 8% below its 20-day simple moving average of $398.16 and 5.8% below its 50-day at $389.14. Both of those averages are declining, which is the signature of a short-term downtrend rather than a pullback.

The 200-day at $369.20 is different. The stock was 0.8% beneath it Thursday and Friday's premarket at $367.06 puts it within 0.6%. That average is the gauge systematic strategies use for the longer-term trend, and the stock has been oscillating around it for days.

The structure that broke was a multi-week sideways range on the daily chart, and the stock has been trading below the high-volume level ever since — a gamma configuration that tends to amplify moves in both directions.

Momentum has been weak throughout the decline. There has been no capitulation candle, no volume spike, no obvious exhaustion — just persistent distribution across seven sessions that took 12% off the price.

One longer-term framework reads the current move as routine. Broadcom has traded within a consistent channel since August 2009, making standard 30% declines when it hits a high. Having made a higher high in June, that framework expects a decline of at least 30.5% — which from the $495.00 peak implies roughly $344.

That target sits 6.3% below Friday's price and is the level where the technical bear case and the fundamental bull case collide.

Institutional flow has been mixed. Recent filings show major institutional buying including additions by Ark Invest, and the earlier characterization of the sideways range was as a breakout and retest with buy continuation expected.

For the near term, the level structure is simple. The 200-day at $369.20 is the pivot. Above it, the 50-day at $389.14 becomes the objective and the 20-day at $398.16 sits behind. Below it, the recent lows and then $344.

Friday's premarket gain took the stock through the pivot. Holding it into the September 2 print is the requirement.

The Levels: $398 Overhead, $358 And $286 Underneath

Immediate resistance. $368.00 is Friday's session high. The 200-day SMA at $369.20 sits immediately above and is the first genuine line — reclaiming it on a daily close changes the longer-term trend classification.

Primary resistance. The 50-day SMA at $389.14 is the next tier, followed closely by the 20-day at $398.16. Both are declining, which means they will meet a rising price rather than the price having to climb all the way to them. Clearing $398 would put the short-term structure back in alignment.

Above that. The $420 area is where the stock traded before the drawdown accelerated. Beyond it, the 52-week high at $495.00 requires a 34.9% advance and would need the September 2 report to deliver a raise to the FY27 AI target.

First support. $360.55 is Friday's session low. The $358 area has been flagged as the key technical zone that traders positioned around during the pullback.

Second support. $344 is the level implied by the historical 30.5% correction pattern from the June high, and it is the reference the technical bears are working toward.

Deeper. Below $344, there is limited structure until the March lows. The 52-week low at $286.17 is 22.0% below Friday's price and would require a genuine AI capex reset rather than a multiple compression.

The framing into September 2: above $369.20 the bias improves with $389 and $398 the objectives. Between $358 and $369 is the current battleground. Below $358 the correction extends toward $344.

The options positioning around the print will be the tell in the final week. Broadcom's history around earnings is violent — the stock fell more than 12% on the June report and had risen 11% in extended trading on a prior beat.

Consensus targets sit far above every one of these levels. The Buy rating carries an average forecast of $513.68 in one survey and $493.24 in another, with 42 analysts averaging a figure roughly 24% above the recent close. A $455 target has also been cited.

That gap between $367 and a $493 to $514 consensus is the widest it has been all year, and it either closes upward on September 2 or the targets come down.

Broadcom Stock Price Forecast: Base, Bull And Bear Into Q4

Base case. The stock consolidates between $358 and $398 into the September 2 report. This is the highest-probability path. Two months of selling have reset expectations, the 200-day at $369.20 is providing a pivot, and the $100 billion financing headline has already been absorbed without a material repricing. A quarter that meets the $29.4 billion revenue guide and delivers $16 billion of AI semiconductor revenue with the FY27 target reiterated would be received as adequate rather than exciting at this valuation. Watch the daily close against $369.20 as the cleanest read on control.

Bull case. A quarter that beats $29.44 billion and, critically, raises the FY27 AI semiconductor target above $100 billion for the first time takes the stock through $398 and opens $420, then the $455 to $495 band. That path requires three things: the $16 billion AI guide being exceeded rather than met, the $100 billion financing being confirmed with terms that preserve investment-grade treatment on the senior tranche, and hyperscaler capex tracking toward the forecast 57% jump in 2027. The Apple agreement exceeding $30 billion through 2031 and new client orders arriving next year support the multi-year case. Consensus targets at $493 to $514 imply 34% to 40% upside from current levels.

Bear case. A third consecutive reiteration of the $100 billion FY27 target without a raise confirms the ceiling read and sends the stock through $358 toward the $344 level implied by the historical correction pattern. The accelerants are specific: Marvell's Google TPU deal eroding the anchor customer relationship, Broadcom's CDS continuing to widen as the market prices $100 billion of contingent guarantees against $19.63 billion of cash, and long-end Treasury yields at 5.25% compressing every multiple in the semiconductor complex. A material change in one large customer's AI spending plan — the concentration risk management itself flags — takes this well below $344.

What actually decides it. Three variables, in order. The FY27 AI semiconductor number on September 2 — raised, reiterated, or trimmed, and nothing else in the release matters as much. The credit spread on the AI XPV paper as roughly $15 billion reaches the secondary market by early 2027, because that is the first public price on what a Broadcom guarantee is worth. And the six core customers, because a business where a handful of names drive the bulk of revenue and all of the growth is only as durable as their capital plans.

At $367.06 the stock has priced two months of disappointment and none of the $100 billion financing. September 2 decides which of those the market was right about.

That's TradingNEWS