Here we go! $DRV
1
60
Happy Birthday SpectreAI!!!! Thank you for grinding during the red candles and bringing us gorgeous UI with smart insights. I've been with you since the beginning and appreciate every update!
Happy Birthday Spectre. 15.11.2023.
Longevity and time in this space matters. Planting seeds, forming formidable communities and pathways that have a plan.
We are young, we are early, we are eager. The growth is ahead of us and it is only just the beginning.
Three years. Thank you for growing with us.
McCoyster retweeted
The @DeriveXYZ RFQ telegram channel is absolutely buzzing today, quite literally my only source of dopamine right now
Some of you wanted another crack at cheap $DRV, here it is. Getting back to silly ratios vs fees
Exactly. Follow the money:
Almanac: 100% of a record $109K in fees used to buy back and burn TAO
Lium: ~5,700 TAO bought back on-chain Nov 12
Targon: 1,772 TAO bought Oct 27
Revenue is the headline. Buybacks are the bid.
Subnets are quietly becoming $TAO's best customers.
Bittensor just reported $2.7 million in GPU rentals. But the smaller $109,440 number might tell us more about the future of $TAO.
Almost everyone is looking at the revenue.
Very few are asking where that money goes.
Lium recorded $2.7M in GPU rentals over 31 days, up 87%. Paying customers increased 72% to 2,394.
That is real commercial traction.
But revenue generated inside the Bittensor ecosystem does not automatically become buying pressure for $TAO.
Now look at Almanac, Subnet 41.
It generated $109,440 in October fees, up 32%, and reported using the entire amount to buy back and permanently burn TAO and subnet alpha.
One subnet is demonstrating that customers will pay for decentralised infrastructure.
Another is demonstrating how customer payments can translate into token demand and reduced supply.
Those are two different achievements. And Bittensor needs both.
Here is the question I think the market is overlooking.
What happens if subnet revenues grow from millions to hundreds of millions, but only a small fraction of that economic activity ever reaches the tokens?
Bittensor could become a successful decentralised AI economy without every token capturing an equal share of that success.
That is why I would start tracking three numbers for every serious subnet:
1. Revenue from actual paying customers.
2. How much of that revenue creates verifiable token demand.
3. How much supply is permanently removed versus newly issued.
The third number matters because burning tokens means little if issuance and selling pressure consistently overwhelm the burn.
Lium is showing that demand exists.
Almanac is showing one possible route for capturing that demand.
Neither proves the entire network has solved value capture.
The next phase of $TAO valuation may depend less on how much AI Bittensor produces and more on how much of that economic value actually reaches the token.
That is the metric I want to see OpenTensor reporting every week.
$TAO down 10% on the week. Meanwhile Lium did $2.7M in GPU rentals in 31 days (+87%), and Almanac spent 100% of a record $109K in fees buying back TAO. Oh, and a miner in Wales is about to run AI on chicken manure. The chart is selling. The customers are buying.
48
Utility they're building on $DRV
There are a few assets out there whose fee generations is consistently up and to the right
Regardless of the price decline
27
Amen! Hungry for some more bite size bullishness $BTC
I think I've done a decent job of giving my community key levels to watch on Bitcoin in real time throughout this bear market
For the most part, those levels have been major inflection points for the market, but I can't predict the future and I don't try to. Thankfully, most of the members at @milkroaddaily are really smart investors and don't expect me to predict daily price action.
Instead of getting dogmatic about what price is going to do, I've been trying to help them plan for whatever it might do
I'm not particularly concerned about anything on the charts right now. In my view. Nothing out of the ordinary has happened
There's a lot to learn from all of this. I'm going to be asking a lot of questions and trying to understand more about many things I don't know
But I do know one very important thing
I'm buying the absolute screaming fuck out of the entire asset class at these prices
And I don't care what anyone else thinks says or does about it
It's not about money. It's about sending a message
Crypto was not created to make you rich
It was created to make you free
This is WARTIME
Believe in somETHing
$BTC $ETH
1
1
78
Perhaps FAA’s Duffy demanding $50 Bullion in emergency funding to overhaul ATC infrastructure last night adds to the spike? Printer goes burrrr…
The blue spike on this Federal Reserve chart shows the exact switch that will ignite the next violent surge in Bitcoin and Gold.
Right now, the Fed's overnight cash cushion (RRPONTSYD) sits at 0.325B 99.99% empty which only tells you the market is compressed ahead of Monday's (16 Nov) Treasury settlement drain.
What I am sharing below are just the 4 basic surface markers.. the real edge lies much deeper in knowing how to read all 4 data points together when they collide over the next 90 days:
1. RPONTSYD (Fed Overnight Repo Injections) jumps off 0.000B:
Look at the Fed's own chart below.. it is sitting completely flat at 0.000B right now after spiking past 31B earlier. The moment year-end funding stress forces that blue line off zero again, liquidity flips from drain to expansion.
2. WTREGEN (Washington's Checking Account) drops from 914.4B: As the Treasury pays out government bills and pushes trapped cash back into commercial banks below 850B, risk assets catch an immediate wave of fuel.
3. DGS10 (10-Year Treasury Yield) breaks back below 5.00%: Gold (~3,883) and Bitcoin (~65.6k) are currently pinned down by 5.15% yields.. the moment bond yields peak and roll under 5.00%, Gold typically breaks out first and Bitcoin follows violently.
4. The US Dollar Index reverses from its 19-month high (103.41): Once Fed repo injections turn on and Treasury cash flows back out, the dollar loses steam releasing the global brake on both assets at once.
Bookmark these 4 data points right now.. anyone can look up a single chart online, yet reading how all 4 liquidity triggers, locking together in real time is a completely different game, and I will keep decoding every shift for you here in simplest possible words!
1
441
Everyday, another bullish catalyst emerges for $TAO. The AI trade may feel broken at the moment but only temporarily so.
The part people are sleeping on: you mint Gamma by BURNING a subnet's alpha. $1 of alpha burned = 1 Gamma of credit, and it's destroyed when you spend it.
So every inference call and compute job takes supply off the market. Usage literally eats supply. 🔥 $TAO
The more I look into Gamma tokens, the more bullish I get on $TAO.
Imagine being able to access and pay for different Bittensor subnet services with just one wallet. No separate accounts or API keys for every subnet.
And this works for AI agents too.
What I really like is that TAO is used to buy subnet alpha tokens, which then get burned into Gamma credits to pay for services.
Thousands of AI agents transacting across subnets 24/7.
Think about that for a second. The potential here is insane.
1
1
95
Hmmm…. Wasn’t it just last year - October 15, 2025 when Binance had a massive $21 Billion flush blamed on a ‘oracle mispricing’?
Binance says it is covering the hack. FALSE. Its traders are paying.
Binance lost ↓$6.36B in a hack on 29 October and says it refilled ↺ 95% of it in four days. In the same weeks traders paid: ↓$2.88B of long positions were force-closed on Binance, and 76% of all liquidations hit longs.
What is left of a liquidated trader's margin goes into Binance's own insurance fund.
Compounded margin calls on other CEXs driven by Binance.
CC: @CryptoWizardd @eliz883 @lukebelmar
1
42
"The Internet reorganized information. Ethereum is reorganizing trust."
Wall Street rushing in tokenizing stocks on Ethereum rails while ETH trades at 4.3% of what it enables. One of those two numbers is wrong, and I don't think it's the $5.7T.
$ETH 💪🏼
Ethereum generates about $5.7 trillion of value for the economy built on it.
The market prices ETH at $246 billion. That’s 4.3% of it.
Today I’m publishing the second edition of my annual Ethereum-as-infrastructure valuation report. Here’s the essence of it.
1/ Ethereum is a public good, so value it like one
Nobody can buy shares in TCP/IP, yet the Internet is the most valuable infrastructure ever built. Its protocols charge almost nothing. Their value shows up everywhere except in their own revenues.
Ethereum is in the same position. Valuing it by its fees, or comparing it to a company, misses most of what it does. Last year I argued it should be valued the way we value the Internet: through what it enables, not what it charges.
2/ Three buckets, like the Internet
The Internet’s value comes in three parts: the companies built on it, the economic activity it carries, and the surplus people get for free. Ethereum maps onto the same three:
a) Ecosystem value: the assets and companies built on Ethereum, at market prices. $127B.
b) Flow value: the economic activity Ethereum settles, mainly stablecoins, plus trading, lending and tokenized assets. $70T a year of dependent flows, worth $3.0T.
c) Trust surplus: what users would pay for trust in the traditional system, minus what they pay on Ethereum. Custody, reconciliation, settlement delays, counterparty risk. Worth $2.6T.
Add a small credit for the EVM standard used by other chains, and Ethereum’s system value comes to about $5.7 trillion.
3/ Every rate is anchored to something real
I didn’t want numbers that look picked to land somewhere. So:
Ethereum is credited by using Visa’s take on the payments it carries, about 0.3%. Visa is the richest settlement network on earth, so that’s a ceiling, not an average.
The trust cost removed is 25 basis points, weighted toward wholesale settlement rather than expensive retail payments.
Annual values are capitalized at 7%, assuming Ethereum’s activity never grows from today. For a network that has grown every year, that’s deliberately conservative.
4/ ETH’s market cap is the benchmark, not part of the total
This is the biggest change from last year. Counting ETH inside the total would double count. So the model compares what Ethereum enables with what the market pays for ETH.
For every $1 the market pays for ETH, Ethereum generates about $23 of value for the ecosystem built on it.
5/ But ETH shouldn’t be worth all of it
Part of that value belongs to others: L2 tokens, DeFi protocols, shares in Robinhood, Circle and Coinbase, a stake in Tether. And public goods never capture everything they enable. That’s the whole point of them.
So the real question is: what share can ETH capture?
The Internet gives us an answer. The companies built on it are worth roughly 10–20% of the economic value it creates. Apply the midpoint, 15%, to Ethereum, and ETH would be worth about $7,000. That’s 3.5× today’s price.
Today the market credits ETH with less than 5%. Under half the low end of the Internet’s range.
6/ The number that’s hardest to argue with
Forget the capitalization multiple entirely. Each year, Ethereum enables about $389 billion of value. That’s 1.6× ETH’s entire market cap.
Ethereum enables more value every year than the market currently pays for ETH in total.
7/ What’s new: a live model you can challenge
This year the report isn’t static. Every figure comes from a live dashboard that refreshes hourly from public sources. You can inspect every input and formula, and use simulation mode to change the assumptions.
I wanted a model anyone could check and challenge. If you disagree with an assumption, change it and see what happens. valuation.ethmrc.com/
8/ What it isn’t
It’s not a price target, and it’s not a forecast. It’s a snapshot of today, measuring the gap between what Ethereum enables and what ETH is credited with. Either ETH’s capture of its own ecosystem rises toward Internet-like levels, or the market is pricing in that it never will.
That’s the debate this framework makes explicit.
The Internet reorganized information. Ethereum is reorganizing trust.
Full report: ethmrc.com/valuing-eth-2026/
PDF: ethmrc.com/valuing-eth-2026.…
Live model: valuation.ethmrc.com/
An expanded version will appear in my upcoming book, TRUSTSHIFT: Ethereum and the Reinvention of Trust. trustshiftbook.com/
1
56
46 trillion won (~$34B) for ONE HBM4E campus. Memory isn't a cycle anymore, it's a construction project. Somewhere a $STX bull is whispering "told you storage was the trade" 🫡
According to Korean media, Samsung is pouring 46 trillion won, about $34 billion, into turning its Onyang campus into its main HBM4E production base.
The new packaging line is scheduled to start running by June 2028, after another expansion at its Cheonan campus finishes by the end of next year.
Samsung also showed off a prototype of its 7th generation HBM4E chip at the groundbreaking with South Korea's President.
According to estimates, Samsung's monthly HBM packaging capacity grows from about 210,000 wafers in 3Q26 to 290,000 by late 2027 and 370,000 by late 2028.
But here is the part I find most interesting because even with all that spending, new capacity barely adds DRAM bits before 2028, because HBM4E and HBM5 dies are so much bigger and need more layers.
Memory makers are spending like crazy and supply STILL can't catch up.
That is why I think this memory cycle lasts much longer than people expect.
38
$BTC tagged the 200-week MA and John's calling no sub-$60K ever again. Base-building season is where tourists leave and the stackers get quiet. 🫡
I don't expect $BTC to go below $60K again
On Wednesday's rollup I said Bitcoin could form a base around $73K and bounce off the 200-day moving average
Or it could go down and slap the 200-week MA, which it now has
The bigger picture: "I do think it's the early phases of a new bull market"
Global demand for liquidity has been rising faster than liquidity itself. When it gets met, "that's bullish for all risk assets, bitcoin particularly"
FT @KyleReidhead @m0xt_ @BitcoinJesusETH
Amen 🙏. NYSE's parent going 24/7 on Uniswap v4 rails with OKX on Nov 23, and $UNI is -11% on the week. Wall St is literally moving into our neighborhood and the market's treating it like a noise complaint. Milk Road stays early 🥛- we need another video
The company that owns the @NYSE is opening a 24/7 tokenized stock venue with OKX on Nov 23, and the trades will run through pools built on @Uniswap v4
$UNI is down about 11% this week like nobody knows this
It runs on X Layer, OKX's Ethereum L2, with more than 70 assets lined up, including Coinbase, Robinhood and Circle stock and 59 Direxion ETFs. @KyleReidhead called this "the bull case for DeFi" on Wednesday's @MilkRoadDaily rollup
When the NYSE's owner builds on Uniswap, I find it hard to be scary beary no matter what the chart looks like this week
Samsung shopping for $MOD the same week $MRVL buys $LITE. Nobody seems to want to rent the picks and shovels anymore, they want to own the hardware. Cooling was "boring" right up until it popped ~29% after hours.
Following the $MRVL and $LITE acquisition announcement…
Samsung Electronics has been doing vertical integration of their own in photonics to thermal:
- New report from Bloomberg states that
Samsung Electronics is in talks to buy $MOD Modine.
Which supplies thermal/cooling infrastructure for AI data centers.
- 2 days ago from Digitimes, Samsung has
purchased series of silicon photonics patents from Rockley Photonics (that previously was bankrupt). Covering modulators and PICs, mainly around design and integration.
This follows earlier leaks that “Samsung is reportedly considering acquisitions in the silicon photonics space”
After Samsung’s reportedly pursues optical engine and PIC manufacturing by 2027, and CPO services by 2029.
We’ll likely see this trend accelerate with M&A vertical integration across AI supply chains.
Especially considering valuations from smaller upstream companies collapsed recently, despite their strategic significance.
73
Respectable honest breakdown. $ONDS doubled sales, beat estimates, raised the year, and pulled profitability forward. Losses widened because they're scaling into an $835M backlog and a $1.1B 2027 target.
Market definitely agrees with the bruised margins - I’m investing in their smart approach to scaling with a deep backlog to justify the spend.
$ONDS Q3 2026 earnings: Drone maker Ondas doubled its revenue, but its losses widened
Ondas, a maker of military drones and anti-drone systems, sold twice as much as last quarter while its losses deepened. Sales from businesses it owns today rose 64% from a year ago. But gross margin dropped to 22%, because new production lines ran inefficiently. Chief executive Eric Brock still lifted the yearly target and set a $1.1 billion goal for next year.
• Growth if today's businesses were owned a year ago: +𝟔𝟒% (plan: more than 30%)
• Revenue: $𝟏𝟔𝟕.𝟓𝐌 (plan: $140–155M)
• Loss before interest, tax and depreciation, excluding one-offs: −$𝟓𝟗.𝟕𝐌 (−$46.2M last quarter)
• Full-year revenue outlook: 𝐑𝐚𝐢𝐬𝐞𝐝 𝐭𝐨 𝐚𝐭 𝐥𝐞𝐚𝐬𝐭 $𝟓𝟓𝟎𝐌 (earlier: $525–550M)
⚖️ 𝐕𝐞𝐫𝐝𝐢𝐜𝐭: 🟢 Bullish
The story got better, because next year's goal rests on fast growth from businesses Ondas already owns. The company also pulled forward its date for a company-wide profit excluding one-offs. One caution: that loss widened after Brock said it would narrow, and Ondas redefined the measure again.
The question now is whether the margin drop reflects a temporary production ramp or the products Ondas now sells. If the ramp, margins recover and the profit dates hold; if the mix, higher sales add little profit. Gross margin early next year will tell, since Ondas expects a similar fourth quarter.
[Model: Claude Opus 5.5. Full article with charts - link in bio]
🐂 𝐁𝐮𝐥𝐥 𝐂𝐚𝐬𝐞
🟢🟢 𝐎𝐧𝐝𝐚𝐬 𝐓𝐚𝐫𝐠𝐞𝐭𝐬 $𝟏.𝟏 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐨𝐟 𝟐𝟎𝟐𝟕 𝐒𝐚𝐥𝐞𝐬
Ondas builds military drones, anti-drone systems and robotic ground vehicles, and it has grown mostly through purchases. Because deals distort any comparison with last year, the company also reports growth as if it had owned today's businesses a year ago. On that basis, sales rose 64%, against a plan of more than 30%.
Revenue reached $167.5 million, about 14% above the middle of the August range, by our math. Chief executive Eric Brock lifted the full-year target to at least $550 million. He also set a first goal of $1.1 billion for next year, "excluding additional acquisitions", with more than 40% growth from businesses Ondas already owns.
But a beat against the plan mainly shows that the plan was cautious. Ondas has topped both quarterly sales ranges it has given, by 21% on average by our math. The new full-year floor sits only $12.5 million above the old middle of the range, so the fourth-quarter goal did not move.
What to watch: fourth-quarter revenue, reported early next year, against the goal of more than $250 million. It will show whether deliveries on the Army's unmanned strike contract began on schedule.
🟢 𝐍𝐞𝐰 𝐎𝐫𝐝𝐞𝐫𝐬 𝐀𝐠𝐚𝐢𝐧 𝐄𝐱𝐜𝐞𝐞𝐝𝐞𝐝 𝐑𝐞𝐯𝐞𝐧𝐮𝐞
Backlog is the value of orders Ondas has signed but not yet delivered. In the third quarter, the drone maker took $214 million of new orders, more than it shipped. Ondas Sentinel, its new US defense unit, booked $100 million of that, including $21.7 million for IonStrike, a product of DZYNE.
Growth in the order book slowed, though. The company says backlog rose 36% to $835 million. That comparison uses a figure that left out DZYNE and Cyberhawk, two businesses Ondas bought in July and August. Against a like-for-like base, the increase was about 10%, by our math.
What to watch: fourth-quarter backlog, which Ondas expects to grow on "strong order visibility". An increase there, while revenue climbs toward its quarterly goal, would show orders still outpacing shipments.
🟢 𝐏𝐨𝐬𝐢𝐭𝐢𝐯𝐞𝐬 𝐭𝐡𝐢𝐬 𝐪𝐮𝐚𝐫𝐭𝐞𝐫 𝐝𝐢𝐝𝐧'𝐭 𝐭𝐞𝐬𝐭
Two strengths in the Ondas case got no new figures in the third-quarter release, so they stand where they stood after the second quarter. The quarterly filing and the call may fill the gaps.
• Software that lets drones, sensors and ground robots work as one system: Brock said Ondas now combines "autonomous systems, mission software, AI, and critical components", but the release gives no software revenue; a first such figure would move this.
• Growth by business: last quarter Ondas gave rates for units such as Sentrycs, its anti-drone business; this release gives only the company-wide rate, and a split would show how widely the growth spreads.
🐻 𝐁𝐞𝐚𝐫 𝐂𝐚𝐬𝐞
🔴🔴 𝐆𝐫𝐨𝐬𝐬 𝐌𝐚𝐫𝐠𝐢𝐧 𝐇𝐚𝐥𝐯𝐞𝐝 𝐚𝐧𝐝 𝐭𝐡𝐞 𝐋𝐨𝐬𝐬 𝐌𝐞𝐚𝐬𝐮𝐫𝐞 𝐂𝐡𝐚𝐧𝐠𝐞𝐝
"The next phase is about disciplined execution: delivering on customer commitments, integrating our capabilities, and converting demand to sustainable growth and improving profitability," Brock said. Yet gross margin, the share of revenue left after the cost of making products, fell to 21.8% from 43.1% in the second quarter.
The cost of goods sold climbed to 78% of revenue from 57%. Ondas blamed amortization of acquired technology and a markup on inventory that came with acquisitions. It also cited "inefficiencies in the early stages of ramping large programs", mostly in precision strike, its attack-drone unit.
Last quarter Brock said the loss excluding one-offs would start to narrow in the third quarter. Instead it widened to $59.7 million, even after a new definition removed $17.1 million of costs.
The release says the change "did not affect previously reported non-GAAP measures". Yet Ondas now shows a second-quarter loss of $46.2 million, against $50.6 million reported in August. So the measure behind its profit goals leaves out more costs each quarter, while the loss on any basis grew.
What to watch: gross margin early next year, after a fourth quarter Ondas expects to match this one. A return toward 43% would show the drop came from the production ramp, not from the product mix.
🔴 𝐒𝐡𝐚𝐫𝐞𝐬 𝐎𝐮𝐭𝐬𝐭𝐚𝐧𝐝𝐢𝐧𝐠 𝐑𝐨𝐬𝐞 𝟓𝟓% 𝐒𝐢𝐧𝐜𝐞 𝐃𝐞𝐜𝐞𝐦𝐛𝐞𝐫
Ondas pays for much of its expansion with new shares, which cuts each owner's portion of future profit. It had 590 million shares at the end of September, against 381 million in December. Stock-based pay, a non-cash cost, equalled 36% of revenue.
Ondas has also promised 45 million shares to DZYNE's sellers in January. Warrants, rights that let investors buy shares at a set price, sit on the balance sheet as a $918.6 million liability. A $125.1 million drop in their value counted as a gain, and that gain shrank the net loss.
What to watch: the share count at the end of the first quarter of next year, once the DZYNE shares arrive. Each further deal paid in stock would add to it.
🔴 𝐂𝐚𝐬𝐡 𝐅𝐞𝐥𝐥 $𝟓𝟕𝟑 𝐌𝐢𝐥𝐥𝐢𝐨𝐧 𝐚𝐬 𝐎𝐧𝐝𝐚𝐬 𝐊𝐞𝐩𝐭 𝐁𝐮𝐲𝐢𝐧𝐠
Ondas has grown by buying companies, and it closed six more deals this quarter, including DZYNE, its largest so far. The company says the purchases widen what it can sell to the US military. It also plans to buy Stemme, maker of the ULTRA S12 airframe, to "strengthen Ondas' production capacity and supply chain".
Cash, restricted cash and short-term investments fell to $820.3 million from $1.39 billion at the end of June. Ondas also owes sellers $268.2 million in fixed and conditional payments. Deal costs came to $14.1 million, and the release ties part of higher research spending to "intensified integration efforts".
What to watch: year-end cash against fourth-quarter revenue. Another drop of this size before Ondas reaches its profit goals would mean it needs to raise more money from shareholders.
🔴 𝐑𝐢𝐬𝐤𝐬 𝐭𝐡𝐢𝐬 𝐪𝐮𝐚𝐫𝐭𝐞𝐫 𝐝𝐢𝐝𝐧'𝐭 𝐚𝐧𝐬𝐰𝐞𝐫
Five standing risks got no new figures in the release. The quarterly filing and the call should answer most of them before the year ends.
• Fourth-quarter load: the plan still needs a steep step-up in revenue, resting on deliveries that start in the fourth quarter; the fourth-quarter report settles it.
• One large customer: a single buyer took 40% of revenue last quarter; the filing's customer table will show whether that share fell.
• Israel: much of Ondas' production sits in a conflict region, and the release gives no update.
• Acquired units losing money: World View and Bird lost money after their purchase; the release gives no results by unit.
• Backlog timing: Ondas gives no schedule for turning backlog into revenue, and no win rate for its pipeline.
👓 𝐎𝐭𝐡𝐞𝐫 𝐓𝐡𝐞𝐦𝐞𝐬
𝐎𝐧𝐝𝐚𝐬 𝐂𝐫𝐞𝐚𝐭𝐞𝐝 𝐒𝐞𝐧𝐭𝐢𝐧𝐞𝐥 𝐚𝐧𝐝 𝐑𝐞𝐠𝐧𝐚 𝐀𝐈 𝐔𝐧𝐢𝐭𝐬
Ondas grouped DZYNE and World View into Ondas Sentinel, a US defense unit meant "to accelerate the development, production and deployment of advanced defense technologies". It also formed Regna AI to build "mission-critical sovereign AI applications", meaning AI that governments run under their own control. Regna "will leverage NVIDIA's Nemotron open-weight AI models and Palantir's integration and deployment platforms". The release gives revenue for neither unit.
𝐂𝐮𝐫𝐫𝐞𝐧𝐜𝐲 𝐌𝐨𝐯𝐞𝐬 𝐂𝐮𝐭 𝐄𝐪𝐮𝐢𝐭𝐲 𝐛𝐲 $𝟏𝟒.𝟔 𝐌𝐢𝐥𝐥𝐢𝐨𝐧
Most of Ondas' backlog sat outside the US at the end of June, so exchange rates now affect its accounts. Converting foreign units' results into dollars cost $14.6 million this quarter. Ondas booked that loss in equity rather than in the income statement, so it did not change the reported net loss. The release names no currency.
— • — • —
💲 𝐎𝐭𝐡𝐞𝐫 𝐊𝐏𝐈𝐬
𝐈𝐧𝐯𝐞𝐧𝐭𝐨𝐫𝐲 (𝟐𝟔𝐐𝟑): $𝟏𝟏𝟎.𝟕 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇗ accelerating
Inventory more than doubled from June, keeping pace with revenue. It equalled 66% of quarterly sales, after 62% in the second quarter. At the end of 2025 the ratio stood at 73%. Part of the jump came with acquired businesses, and the stock ties up cash while Ondas still loses money.
𝐑𝐞𝐜𝐞𝐢𝐯𝐚𝐛𝐥𝐞𝐬 (𝟐𝟔𝐐𝟑): $𝟏𝟐𝟕.𝟑 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇗ accelerating
Receivables, money customers owe for goods already delivered, stood at 76% of quarterly sales. That sits below the 90% of the first quarter, so collections have kept up with faster sales. The balance has climbed with revenue and with the businesses Ondas bought.
𝐃𝐞𝐟𝐞𝐫𝐫𝐞𝐝 𝐫𝐞𝐯𝐞𝐧𝐮𝐞 (𝟐𝟔𝐐𝟑): $𝟑𝟕.𝟖 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇘ decelerating
Deferred revenue, cash customers paid before delivery, grew 41% from June, less than half the pace of sales. It has climbed every quarter this year, from $8.0 million in December. Because it lags sales, advance payments now fund a smaller share of the drone maker's production.
𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐚𝐧𝐝 𝐝𝐞𝐯𝐞𝐥𝐨𝐩𝐦𝐞𝐧𝐭 (𝟐𝟔𝐐𝟑): $𝟔𝟏.𝟔 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇒ stable
Research spending doubled from the second quarter, in step with revenue, so it held at about 37% of sales. The release ties the rise to faster product development and integration work across acquired firms. At this level, research alone costs more than the quarter's entire gross profit.
— • — • —
🔮 𝐆𝐮𝐢𝐝𝐚𝐧𝐜𝐞
𝐅𝐘𝟐𝟔 𝐑𝐞𝐯𝐞𝐧𝐮𝐞: 𝐀𝐭 𝐥𝐞𝐚𝐬𝐭 $𝟓𝟓𝟎 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇗ accelerating
🠅 raised from $525–550 million
Raised. The plan now calls for sales more than ten times those of 2025. By our math, the first nine months leave about $249 million for the fourth quarter, just under the separate quarterly goal. Fifth increase in five quarters: +$265M, +$15M, +$147.5M, then +$12.5M against the old middle of the range.
𝐐𝟒 𝐑𝐞𝐯𝐞𝐧𝐮𝐞: 𝐌𝐨𝐫𝐞 𝐭𝐡𝐚𝐧 $𝟐𝟓𝟎 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇗ accelerating
🠆 unchanged from About $236–276 million implied by the full-year range; a $1 billion annual pace exiting 2026
Unchanged. The forecast asks for about 49% more revenue than in the third quarter. That is more than eight times the year-ago quarter. Ondas expects volume deliveries on the Army's unmanned strike contract and its combat engineering vehicle programme to carry the step-up. Brock last quarter tied the same level to a $1 billion yearly sales pace.
𝐐𝟒 𝐏𝐫𝐨 𝐟𝐨𝐫𝐦𝐚 𝐨𝐫𝐠𝐚𝐧𝐢𝐜 𝐫𝐞𝐯𝐞𝐧𝐮𝐞 𝐠𝐫𝐨𝐰𝐭𝐡: 𝐀𝐭 𝐥𝐞𝐚𝐬𝐭 𝟕𝟓% ⇗ accelerating
New. Ondas expects sales from today's businesses to rise at least 75% from a year earlier, after 64% in the third quarter. By our math, that caps the comparable year-ago base at about $143 million. The company calls the expected growth "broad-based across Ondas' product portfolio".
𝐅𝐘𝟐𝟕 𝐑𝐞𝐯𝐞𝐧𝐮𝐞: 𝐀𝐭 𝐥𝐞𝐚𝐬𝐭 $𝟏.𝟏 𝐛𝐢𝐥𝐥𝐢𝐨𝐧, 𝐰𝐢𝐭𝐡 𝐦𝐨𝐫𝐞 𝐭𝐡𝐚𝐧 𝟒𝟎% 𝐨𝐫𝐠𝐚𝐧𝐢𝐜 𝐠𝐫𝐨𝐰𝐭𝐡, 𝐞𝐱𝐜𝐥𝐮𝐝𝐢𝐧𝐠 𝐚𝐝𝐝𝐢𝐭𝐢𝐨𝐧𝐚𝐥 𝐚𝐜𝐪𝐮𝐢𝐬𝐢𝐭𝐢𝐨𝐧𝐬 ⇘ decelerating
New. The goal doubles the 2026 floor without further deals and assumes more than 40% growth from today's businesses. By our math, those businesses would then have earned about $786 million this year under full ownership. That rate sits well below the 75% Ondas expects for the fourth quarter.
𝐐𝟒 𝐁𝐚𝐜𝐤𝐥𝐨𝐠: 𝐒𝐞𝐪𝐮𝐞𝐧𝐭𝐢𝐚𝐥 𝐠𝐫𝐨𝐰𝐭𝐡 𝐟𝐫𝐨𝐦 $𝟖𝟑𝟓 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 ⇒ stable
New. Ondas expects backlog to rise again in the fourth quarter. With quarterly revenue planned above $250 million, new orders must at least match that, by our math. That would top every quarter of orders the company has reported.
𝐐𝟒 𝐆𝐫𝐨𝐬𝐬 𝐦𝐚𝐫𝐠𝐢𝐧: 𝐒𝐢𝐦𝐢𝐥𝐚𝐫 𝐭𝐨 𝐐𝟑 (𝟐𝟏.𝟖% 𝐫𝐞𝐩𝐨𝐫𝐭𝐞𝐝; 𝟑𝟒.𝟖% 𝐞𝐱𝐜𝐥𝐮𝐝𝐢𝐧𝐠 𝐨𝐧𝐞-𝐨𝐟𝐟𝐬) ⇒ stable
New. The company expects "similar gross margin levels in Q4 2026 due to mix", then a return to historical levels. Over the three quarters before this one, reported gross margin ran between 42% and 49%. The plan gives no date for the recovery.
𝐀𝐝𝐣𝐮𝐬𝐭𝐞𝐝 𝐄𝐁𝐈𝐓𝐃𝐀 𝐥𝐨𝐬𝐬, 𝐜𝐨𝐦𝐢𝐧𝐠 𝐪𝐮𝐚𝐫𝐭𝐞𝐫𝐬: 𝐒𝐢𝐠𝐧𝐢𝐟𝐢𝐜𝐚𝐧𝐭 𝐝𝐞𝐜𝐥𝐢𝐧𝐞𝐬 𝐢𝐧 𝐥𝐨𝐬𝐬𝐞𝐬 𝐢𝐧 𝐭𝐡𝐞 𝐜𝐨𝐦𝐢𝐧𝐠 𝐪𝐮𝐚𝐫𝐭𝐞𝐫𝐬 ⇄ reversing
🠇 cut from Losses to narrow in the second half, beginning in Q3
Pushed back. Brock had said the loss excluding one-offs would start to narrow in the third quarter, but it widened. The release now promises "significant declines" in losses over the coming quarters. It calls this year's spending "a front-loading of expenses ahead of the continued significant revenue ramp into 2027".
𝐎𝐩𝐞𝐫𝐚𝐭𝐢𝐧𝐠-𝐩𝐥𝐚𝐭𝐟𝐨𝐫𝐦 𝐚𝐝𝐣𝐮𝐬𝐭𝐞𝐝 𝐄𝐁𝐈𝐓𝐃𝐀 𝐩𝐫𝐨𝐟𝐢𝐭: 𝐐𝟒 𝟐𝟎𝟐𝟔 (𝐎𝐧𝐝𝐚𝐬 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 𝐚𝐧𝐝 𝐎𝐧𝐝𝐚𝐬 𝐒𝐞𝐧𝐭𝐢𝐧𝐞𝐥) ⇄ reversing
🠆 unchanged from Q4 2026
Unchanged. Ondas still expects its operating businesses, Ondas Autonomous Systems and Ondas Sentinel, to earn a profit excluding one-offs in the fourth quarter. The goal leaves out head-office and deal costs. The release gives no third-quarter figure for this level, so readers cannot yet measure the gap.
𝐂𝐨𝐦𝐩𝐚𝐧𝐲-𝐰𝐢𝐝𝐞 𝐚𝐝𝐣𝐮𝐬𝐭𝐞𝐝 𝐄𝐁𝐈𝐓𝐃𝐀 𝐩𝐫𝐨𝐟𝐢𝐭: 𝐁𝐲 𝐐𝟑 𝟐𝟎𝟐𝟕 ⇄ reversing
🠅 raised from Q4 2027
Pulled forward. Ondas moved the date "one quarter earlier than its prior expectations", citing revenue and gross profit growth against slower expense growth. The date has moved earlier in each of the last two quarters. The goal rests on a measure Ondas redefined again this quarter.
𝐂𝐨𝐦𝐛𝐚𝐭 𝐞𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐢𝐧𝐠 𝐯𝐞𝐡𝐢𝐜𝐥𝐞 𝐝𝐞𝐥𝐢𝐯𝐞𝐫𝐢𝐞𝐬: 𝐕𝐨𝐥𝐮𝐦𝐞 𝐝𝐞𝐥𝐢𝐯𝐞𝐫𝐢𝐞𝐬 𝐛𝐞𝐠𝐢𝐧 𝐢𝐧 𝐐𝟒 𝟐𝟎𝟐𝟔 ($𝟏𝟒𝟎 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 𝐩𝐫𝐨𝐠𝐫𝐚𝐦) ⇗ accelerating
🠆 unchanged from Volume deliveries expected to begin in Q4 2026
Unchanged. Ondas still plans volume deliveries this quarter on the $140 million combat engineering vehicle programme, run by Indo Earth, its earth-moving equipment unit. The fourth-quarter revenue goal leans on this start.
𝐀𝐫𝐦𝐲 𝐋𝐞𝐭𝐡𝐚𝐥 𝐔𝐧𝐦𝐚𝐧𝐧𝐞𝐝 𝐒𝐭𝐫𝐢𝐤𝐞 𝐬𝐡𝐢𝐩𝐦𝐞𝐧𝐭𝐬: 𝐂𝐨𝐧𝐭𝐢𝐧𝐮𝐞𝐝 𝐯𝐨𝐥𝐮𝐦𝐞 𝐬𝐡𝐢𝐩𝐦𝐞𝐧𝐭𝐬 𝐮𝐧𝐝𝐞𝐫 𝐭𝐡𝐞 $𝟗𝟖𝟐 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 𝐔.𝐒. 𝐀𝐫𝐦𝐲 𝐈𝐃𝐈𝐐 ⇗ accelerating
🠆 unchanged from Production ramp starting in Q3 2026
Unchanged. Ondas expects continued volume shipments by Mistral, its US unit, under the Army's $982 million Lethal Unmanned Strike contract. That contract is a multi-year framework that sets a ceiling on orders. The release gives no shipment figure for the third quarter.
— • — • —
❓ 𝐊𝐞𝐲 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧𝐬
𝐖𝐡𝐲 𝐝𝐢𝐝 𝐥𝐚𝐬𝐭 𝐪𝐮𝐚𝐫𝐭𝐞𝐫'𝐬 𝐥𝐨𝐬𝐬 𝐟𝐢𝐠𝐮𝐫𝐞 𝐜𝐡𝐚𝐧𝐠𝐞?
The release says the new exclusions did not touch past figures, yet the second-quarter loss excluding one-offs moved from $50.6 million to $46.2 million. Readers need one consistent series to judge the profit dates.
𝐖𝐡𝐢𝐜𝐡 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐬 𝐩𝐮𝐥𝐥𝐞𝐝 𝐠𝐫𝐨𝐬𝐬 𝐦𝐚𝐫𝐠𝐢𝐧 𝐝𝐨𝐰𝐧, 𝐚𝐧𝐝 𝐡𝐨𝐰 𝐟𝐚𝐫?
Ondas named precision strike but gave no margin by unit. A split would show whether the low margin expected for the fourth quarter comes from a few ramping programmes or from the product mix itself.
𝐇𝐨𝐰 𝐦𝐮𝐜𝐡 𝐛𝐚𝐜𝐤𝐥𝐨𝐠 𝐛𝐞𝐜𝐨𝐦𝐞𝐬 𝐫𝐞𝐯𝐞𝐧𝐮𝐞 𝐰𝐢𝐭𝐡𝐢𝐧 𝐭𝐰𝐞𝐥𝐯𝐞 𝐦𝐨𝐧𝐭𝐡𝐬?
Ondas gives no conversion schedule for its backlog. The share due within a year, and how much of the fourth quarter rests on the Army strike contract, would show how secure the revenue goal is.
𝐃𝐨𝐞𝐬 𝐨𝐧𝐞 𝐜𝐮𝐬𝐭𝐨𝐦𝐞𝐫 𝐬𝐭𝐢𝐥𝐥 𝐩𝐫𝐨𝐯𝐢𝐝𝐞 𝟒𝟎% 𝐨𝐟 𝐫𝐞𝐯𝐞𝐧𝐮𝐞?
Last quarter a single buyer took 40% of revenue and 43% of receivables. The release names neither the customer nor its programme; the filing will show whether acquisitions have diluted that share.
𝐖𝐡𝐚𝐭 𝐝𝐢𝐝 𝐃𝐙𝐘𝐍𝐄 𝐚𝐧𝐝 𝐂𝐲𝐛𝐞𝐫𝐡𝐚𝐰𝐤 𝐚𝐝𝐝 𝐭𝐨 𝐫𝐞𝐯𝐞𝐧𝐮𝐞?
Both joined during the quarter. Their revenue, and the DZYNE figures the quarterly filing must report, would separate bought growth from the rate Ondas gives for the businesses it owns today.
























