Sponsored

edgeful: Back your trades with real data, not just hunches.

We've all had those moments, spotting what looks like a solid setup, but wondering if it's backed by history or just wishful thinking.

If you want to add some straightforward stats to your process without drowning in complexity, edgeful could be a solid option. I've checked it out myself lately, and it's given me quick insights into how markets have played out in similar spots.

What edgeful offers is a clean way to dig into things like price action patterns, trend behaviors, and even performance on specific weekdays. It's not about magic forecasts; it's more about pulling historical data to show probabilities, so you can line them up with current conditions.

For example, on stocks, it might break down how often a certain breakout level has held or reversed based on past sessions. In futures or forex, you could check volume trends or gap behaviors to gauge if a move has legs. It's practical stuff that helps confirm if a trade fits the odds or if it's smarter to wait.

The tool covers stocks, futures, forex, and crypto, making it versatile if you trade across different areas.

Heads up for my readers: Edgeful doesn't usually offer free trials, but they've hooked us up with a 7-day one. If it sounds useful, check it out via the link below—no strings attached, and sometimes passing is the best call.

👉 Grab the 7-day edgeful trial here.

Two lanes, one race:

Broadcom just gave the AI hardware story a new plot line. The company said it has a new custom-chip customer with about $10 billion of orders starting in fiscal 2H26, and multiple outlets report that the customer is OpenAI. The Financial Times says the co-designed processor is slated to move into mass production in 2026, with Reuters echoing that timeline. If that holds, Broadcom’s AI semiconductor revenue could run “well over” $40 billion next year, versus a roughly $30 billion view only a quarter ago.

This sits on top of a solid quarter. Broadcom’s AI semi revenue was about $5.2 billion in the July quarter, up 63% year on year, and management guided the October quarter to $6.2 billion. Total Q4 revenue guidance is $17.4 billion, and the company highlighted a record backlog north of $100 billion. That is the backdrop for the OpenAI news and for management’s message that AI growth “accelerates” in FY26.

So what does an OpenAI ASIC mean in practice, and how should we think about Nvidia in the same breath? An ASIC is a purpose-built chip. If you run the same workload millions of times, a custom part can be cheaper, faster, and more power-efficient than a general GPU. It is not a full swap for everything. Training frontier models, trying novel architectures, and serving a changing mix of tasks still benefit from Nvidia’s programmable platform. Even Nvidia’s CEO has been blunt that many ASIC projects start and few reach production because the systems and software complexity is hard to tame at scale. That is chest-thumping, but it is also a real constraint that slows competitors.

OpenAI’s calculus is straightforward. Its inference bill is exploding. If it can freeze parts of its workload and push them to a Broadcom-built accelerator, unit economics improve and it reduces single-vendor dependence. The FT and Reuters both stress that these chips are intended for OpenAI’s internal use, not commercial sale, which keeps the near-term impact contained to OpenAI’s data centers rather than a broad market shift. If the program hits its milestones, the prize is lower cost per token and a bigger runway to scale usage. If it slips, OpenAI still has Nvidia to lean on.

For Nvidia, this is not an existential threat. It is a nudge that the AI pie is diversifying. Hyperscalers already run mixed estates: Google pairs homegrown TPUs with fleets of Nvidia for cloud customers, and Amazon promotes its own silicon while buying GPUs hand over fist. An OpenAI ASIC follows that pattern. The near-term swing factor for Nvidia remains supply, product cadence, and software lock-in, not a single customer’s ASIC tape-out. The medium-term risk is if more workloads become “frozen” and shift to fixed-function parts faster than Nvidia expands into systems and networking to defend its share. Management’s posture suggests they know this and are racing on systems integration, interconnect, and platform software for exactly that reason.

For Broadcom, the OpenAI ramp validates a multi-year bet on custom compute. It already builds for Alphabet, Meta, and ByteDance; adding a high-velocity AI lab tightens the narrative. The numbers matter, though. Custom programs are lumpy, schedules can slip, and major customers tend to multisource. Even so, the company’s guidance and outside estimates imply FY26 AI revenue well north of earlier targets, which is why the multiple expanded on the print. Delivery in 2026 is the test.

The takeaway is simple. The AI compute market is getting large enough to support both a dominant general-purpose platform and a growing lane for custom silicon.

Homebuilders Catch a Bid

Homebuilder stocks caught a strong tailwind over the last month as rates and yields finally broke lower and policy chatter heated up. The average 30-year mortgage rate slid to about 6.29% on Friday after a softer jobs print, the biggest one-day drop in over a year, according to Mortgage News Daily. That is the lowest since early October and a meaningful shift from the 7% handle we saw in May. Stocks in the group responded in kind.

Treasury Secretary Scott Bessent also has been floated that the administration may declare a national housing emergency this fall. The trial balloon included ideas like standardizing building and zoning codes, trimming closing costs, and exploring tariff exclusions on construction inputs. Markets heard “real supply help might be coming,” even if implementation would take time and face local constraints.

Why lower yields matter so much here

Mortgage rates track the 10-year Treasury plus a spread that compensates lenders and investors. That spread has narrowed to the tightest in three years but is still wider than normal, which means there is room for rates to fall further if the bond rally holds and spreads keep tightening. The “normal” range is roughly 150 to 200 basis points over the 10-year. Late August, the spread was still around 226 basis points.

Mechanically, a falling 10-year and a tighter mortgage spread reduce monthly payments and expand the qualified buyer pool. Lenders are already reporting some quotes in the high-5% range for well-qualified borrowers as rate sheets catch up to the bond move.

Demand, supply, and sentiment

This is still a split market. New-home sales in July were running at a 652,000 annualized pace, with 499,000 homes for sale. Median new-home prices are down year over year, which tells you builders are meeting the market. But sentiment remains muted. The NAHB/Wells Fargo Housing Market Index sits near cycle lows, and two-thirds of builders say they are using incentives like rate buydowns. About 37% report outright price cuts, averaging about five percent.

Existing-home activity is still sluggish. July resales ran near a 4.0 million annualized pace as the lock-in effect keeps would-be sellers on the sidelines. That keeps new construction unusually important in meeting demand, especially in fast-growing Sun Belt markets.

What this means for builder P&L

Through the past year, public builders leaned hard on rate buydowns and other incentives to keep sales moving. Lennar spelled it out in June, noting incentives are primarily mortgage buydowns, and third-party research pegged Lennar’s average incentive around 13% of the selling price in Q2, the highest since 2009. D.R. Horton’s latest quarter showed strong orders but margin pressure from incentives. If mortgage rates drift down, builders can dial back the buydown spend first, which helps gross margins before it shows up in price. That is one reason the group is so sensitive to every 10-year move.

Policy watch: upside and caveats

The emergency-declaration talk matters because the bottleneck is not only rates. It is zoning, fees, and input costs. A federal push on model codes, faster permitting, or carrots for local zoning reform could help starts over time. So would tariff relief on construction inputs. NAHB estimates spring tariff actions added roughly nine to eleven thousand dollars to the cost of a typical new home. Rolling back or excluding key materials would be an immediate tailwind to affordability and builder margins. The caveat is that land use is mostly local, and federal levers have limits.

There is also a capital markets angle. Analysts have raised ideas like more aggressive Fed reinvestment in agency MBS or administrative steps involving the GSEs to tighten the mortgage spread. Those are controversial, but they explain why housing headlines have been moving the space even independent of the Fed funds path.

Where we are in the cycle

Production is stabilizing rather than booming. July housing starts ticked higher to a 1.43 million annualized pace, with single-family at 939,000 and multifamily rebounding. New-home sales are steady but not hot. This looks like an early-cycle turn for builders rather than a demand surge. If the rate move holds and spreads tighten, incentives can fade and margins heal before volumes truly expand. If rates back up, incentives likely rise again, and margin relief gets delayed.

Breakup Averted

Alphabet just cleared the ugliest part of its search antitrust overhang. Judge Amit Mehta declined to force a breakup or a sale of Chrome or Android. He did, however, bar exclusive distribution deals and ordered Alphabet to share some search data with qualified rivals. Payments for default placement can continue so long as they are not exclusive.

Let me breakdown:

1) Distribution is intact, just more contested.
The core flywheel of owning Chrome, steering defaults, and paying partners lives on, only without exclusivity. Practically, default status will remain valuable, and manufacturers still like the checks. The difference now is that bidders for default can get a shot, and Google will have to win on price and product, not lockout. Expect more auctions for default placement and shorter deal tenors.

2) Data sharing is a managed risk, not a cliff.
This is not an open-sourcing of Google Search. The court narrowed the obligation to defined search index and interaction data for qualified rivals, and it does not require Google to hand over ad data or hand the keys to its ranking systems. There is competitive help here for challengers, but the moat is still deep: scale infrastructure, spam fighting, freshness, and product integration across Chrome, Android, and YouTube.

3) The battleground shifts to AI assistants and access points.
The order explicitly reaches GenAI access points. That means no exclusive pushes for Gemini as the default AI on devices. The next fight is who controls the on-ramp for AI answers inside the browser, the phone, and the car. Keeping Chrome matters because agents will increasingly live where people browse, shop, and watch. Losing Chrome would have crippled that plan. Keeping it is the difference between offense and defense.

4) The Apple question is now about price and product, not permission.
The court did not ban revenue-share payments to Apple, and that did get reflected in the share price the following day. However, It did ban exclusivity. That keeps a massive distribution channel open while injecting competitive tension. In practice, Apple will still choose the best default experience for its users, and today that is usually Google, but the door is open for alternatives on specific surfaces or regions.

5) Policy risk is not gone, it is normalized.
The six-year term, oversight, and the prospect of appeal keep headlines alive, but they are contextual risks, not existential ones. The judge pointed to AI as a fast-moving market and rejected remedies that would freeze innovation. The ad-tech case is still out there, though the search ruling shows a clear preference for behavioral fixes over breakups.

After this ruling, the biggest overhang on Alphabet has lifted. And in the same breath, Gemini continues to look like the best AI model in market tests. Reports that Apple is in talks with Google to integrate Gemini only reinforce that the skew remains positive for Alphabet going forward.

TSMC’s China Waiver Pulled:

This week Bloomberg reported that the U.S. will revoke TSMC’s waiver that allowed seamless shipments of tools and materials to its Nanjing fab. Starting Dec. 31, 2025, every spare part and piece of gear will require individual U.S. export licenses, ending the blanket authorization known as “validated end user” (VEU). Shares of TSMC and suppliers like Tokyo Electron dipped on the news.

The move aligns TSMC with Samsung and SK Hynix, which lost the same status last week. Washington says licenses will still be issued to keep these fabs running, but thousands of extra permits a year mean delays, bureaucracy, and uncertainty in day-to-day operations. For TSMC, the direct financial hit looks limited—Nanjing is older-node, about 3% of global capacity, but any friction in tool swaps or chemical supplies complicates fab planning.

Robinhood Finally Joins the S&P 500

After years of waiting, Robinhood (HOOD) is being added to the S&P 500, effective September 22. The move comes alongside AppLovin (APP) and Emcor Group (EME), replacing Enphase (ENPH), Caesars (CZR), and MarketAxess ($MKTX). Uber (UBER) will also join the S&P 100.

For Robinhood, this is a milestone. The company went public in 2021 but was repeatedly left out of prior rebalancings, frustrating both management and investors. Profitability hurdles and valuation swings kept it on the sidelines until now. With a market cap north of $20B and steady earnings momentum, the index committee finally gave it the nod.

Inclusion matters because passive funds that track the S&P 500 are now required buyers, creating a built-in wave of institutional demand. That helps explain why shares jumped ~6% after the announcement. Beyond the near-term stock reaction, it also signals that Robinhood has moved past its “meme stock era” identity and is now treated as part of the U.S. equity market’s core benchmark.

Week Ahead:

Markets head into the second week of September carrying the weight of a weak jobs print and a stretched valuation backdrop. The S&P 500 is trading above 24x forward earnings, which makes incoming macro data especially important for rate expectations.

On Wednesday, the Producer Price Index lands, followed by Thursday’s CPI. These two reports will shape the Fed’s September 17 meeting and its updated economic projections. A cooler set of prints could cement the case for multiple cuts this year, but September’s reputation as a difficult month means even “good” data may not guarantee follow-through.

Apple’s iPhone 17 reveal on Tuesday at 10am ET, will dominate headlines, with investors watching not just hardware updates but any AI or “Apple Intelligence” integration. Oracle also reports Tuesday, where cloud growth will be under the microscope, and AeroVironment’s update comes against a favorable defense-spending backdrop. GameStop is also on deck, though expectations are muted. Later in the week, Kroger (Thursday) offers a read on food inflation and consumer trade-downs, while Adobe faces questions about AI competition despite its entrenched position in creative software.

President Trump has flagged “substantial” semiconductor tariffs as imminent, with exemptions for firms that localize U.S. manufacturing — including Apple. Any confirmation or detail here could ripple through the chip supply chain. Abroad, the ECB meets Thursday and is expected to hold rates, while CPI data across Europe and Asia will round out the week.

Should be an interesting week. I’ll be tracking it all on X (formerly Twitter): @wallstengine.