Here’s a preview of what we’ll cover this week: 

Macro: Stop Trading The Payroll Report

Markets: What Do Semis Earnings Tell Us?; What Are We Buying?; Semis Are Back; Nvidia's Art Of War; What to Look For in Anthropic’s S-1?; The Constraint Was Never Technology

Lumida Curations: Rene Haas On AI Demand and Supply; Sam Altman on OpenAI’s IPO; Gavin Baker on Data Centers And Working-Class America

I did an FSD livestream this week, titled “Are Semis Back?”.

Here’s what I talk about:

• What to look for in the Anthropic S-1
• Investing in electric grid infra
• China value
• Post-war Ukraine
• The AI silicon race
• Schumpeter and creative destruction
• Leadership in the Age of AI
• Raising kids in the age of AI

Watch the stream here.

We will be hosting a community get-together on September 16th in New York. Limited spots available.

We would love to host you in an evening of meaningful discussions around equities, private markets, and latest AI opportunities.

Drop a reply to the ledger email to let us know if you can join.

AI tends toward consensus.

Think about what a model does.

A model is essentially multiplying matrices to produce a probability distribution over possible outputs.

Run the same prompt many times and you will get a range of answers.

Most of these answers will cluster around the center of the distribution.

So AI will tend toward consensus. Not because it is badly trained, but because that is what the architecture is for.

The caveat?

Investment returns do not live in the center of the distribution, or more simply, where consensus is. They live in the tails.

Consider Elon. “I am going to build reusable rockets, and simultaneously I am going to build an electric car company. These markets don’t exist yet, and are small, but they will be big. And, I’ll try to connect people brains to machines via a neural link interface, too”

Capital intensive. Unfocused. Competing with incumbents on multiple fronts - including Google on driverless cars - at once.

Any competent investor would have rejected that business plan.

Many did. So, any LLM model is highly likely to do the same.

(Meanwhile, Sequoia is set to enjoy about $15 Bn+ from its early investment in SpaceX, proving LLMs will get it wrong.)

Leadership matters more than ever in the age of AI. 

This is why I think humans maintain role, and it is a bigger role than people expect.

Humans have better vision, have agency, and can make hard decisions.

And it is remarkably easy to nudge a model off its course.

You can push back on a top-flight AI model, and it crumples. 

There is also a structural gap. You cannot yet attach accountability to an agent.

There is no construct for it. No liability, no reputation, no career consequence.

So what compounds from here?

Leadership. Vision. Storytelling. The ability to attract and organize a world-class team.

AI has not disrupted that. AI has enabled these high-agency founders and leaders to create much more impact.

So, what does it mean for kids?

If you have kids, teach them entrepreneurship.

Entrepreneurship is the summary statement for leadership, storytelling, problem solving, critical thinking, curiosity, resilience, grit, and growth mindset in Carol Dweck's sense.

And, maybe the most direct way to teach them entrepreneurship is to start a business with your kids.

Approach it as learning and play rather than performance and you'll get further.

Go build something with your kids.

That may be the best life advantage you can give them.

We are currently working on a new service that uses AI to identify the best tax mitigation opportunities for your unique situation.

Please try it out at Tax.Lumida.com and give us your feedback.

If you like it, we’ll roll it out to the public and finish up the paywalls.

Right now the tax mitigation detection tool (for US citizens only) is free.

Note: Lumida has a wide range of strategies designed to mitigate capital gains tax, estate tax, and income tax. Ping [email protected] if you’d like to learn more.

The most interesting AI deployment I heard about this week was at a lumber mill.

A friend runs a lumber mill.

He is using Grokbot for procurement, for pricing negotiations, and for monitoring inventory levels and controls. He is creating real value from it. 

I have spent the last 72 hours spinning up a couple dozen agents on Grok Bot.

I have an agent that prepares outreach emails. I give it a name, and it reads the bio, drafts the note, attaches prior guest links and the PDF, finds the right contacts, and notifies my team in Slack. 

I have an agent that screens resumes. Another agent that DMs recruits on LinkedIn

I have an agent that pings our marketing team if an FSD video is not published by 8am the next day.

I have an agent that is collecting K-1s and coordinating with my CPA and bookkeeper. (We are going to build this for our Lumida clients…)

I have an agent that is an Executive Assistant - reads my emails, inbound DMs on Slack, Telegram, and Linked In.

I have another agent that scans headlines for me each day across major newspapers. (We already put this in the Lumida Invest App along with myriad other investing bots.)

I have an agent that looks for podcast guests and reaches out to them.

And I have an agent that transcribes every FSD video, reads the Telegram chat where I drop charts, reads my posts, and proofreads the first draft of this newsletter.

We have about a dozen agents at Lumida now. I'd guess that number is in the hundreds within two or three months.

An extraordinary number of workflows can be delegated today, but we don't recognize them as workflows because we take them for granted. 

It is a failure of imagination rather than capability.

The agent I actually want next is a meta-agent: one that says Ram, based on your context, here are the agents you should build, and here are the questions you should be asking. 

All to say, the productivity revolution from AI is very real.

Learn Grok Bot this weekend. Align your human and financial capital against the secular trend of AI.

We’ll talk about semis below, which we now find attractive.

We got the August NFP on Friday, coming in at 162K against 55K expected.

So much for all the fretting about AI job displacement.

Unemployment held at 4.1%, in line. 

Markets traded NFP under ‘Good news is bad news’, as markets are concerned the FOMC will raise rates this month.

Hike odds for September ran up to 60%, having slipped below 50% earlier after Christopher Waller (Fed Governor) spoke on Thursday.

I am not a fan of the non-farm payrolls report.

The confidence intervals are enormous. The revisions are large. The sampling error alone is wide enough to swallow the number the market reacts to.

And the economy does not turn on a dime.

It is a dynamic system that is constantly refactoring itself. It does not deliver its message once a month at 8:30 in the morning.

You can see the state of the economy far more clearly in company earnings. 

Corporates report revenue they actually collected from customers who actually paid, across ninety days. That is closer to a census. Payrolls is a survey with a response rate.

So, where are we on rates?

Our base case has been no hike and no cut in 2026, and Waller's Thursday speech was the strongest support for it.

Waller argues the PCE readings are backward-looking and "not the best guide for where inflation is today." 

Warsh isn’t telegraphing to markets anymore, so Waller has naturally filled that spot.

Nature abhors a vacuum.

On the two supply-side scares of this cycle, he says the price effects of tariffs have largely passed through.

His earlier concern that higher energy prices would bleed into broader goods and services has not materialized.

Waller added that wage growth, once you account for productivity, “is broadly consistent with inflation continuing down toward 2 percent.” 

(AI productivity is showing up in numbers, and will become a disinflationary force in the long run.)

Warsh made it clear that he doesn’t trade the NFP, and so did Waller. 

Warsh loves the AI productivity story, too.

We don’t see the FOMC raising rates in September.

That also means markets are overly bearish. Midterm election seasonality hits the nadir in September, and we’d expect markets to fully discount election risks soon.

Markets had one worry this year: Will the AI demand weaken?

The opposite happened. 

AI demand is stronger than ever before, and we are seeing this across the flywheel. 

It’s funny how they used almost similar phrases to describe the demand backdrop.

Lip-Bu Tan (CEO, Intel) opened with "strong demand for our products continuing to outpace our growing supply." 

Jeff Clarke (COO, Dell): "Demand for our solutions is exceeding available supply." 

Hock Tan (CEO, Broadcom) called Q3 demand: "simply hot, and we're just getting started." 

Arkady Volozh (CEO, Nebius) described demand as "enormous."

Four layers of the supply chain, and each of them had almost the same one sentence. 

And, it’s not only about this quarter. Most of these companies see years of growing demand ahead. 

Colette Kress (CFO, Nvidia) guided to "approximately 70%" revenue growth in fiscal 2028, and flagged it as "a supply-constrained outlook" rather than a demand forecast. 

Broadcom went further and pre-sold it: Hock Tan (CEO) noted Broadcom has "secured the supply to again double AI revenue to approximately $115 billion" in 2027, and volunteered that "our demand actually exceeds this outlook."

Applied Materials sits furthest upstream and therefore sees furthest out. 

Brice Hill (CFO, Applied Materials) says customers are giving them "longer visibility than we've ever had with some conversations now extending to 2030."

What Is Driving This Demand?

Jensen Huang (CEO, Nvidia) noted the increase in compute demand is driven by adoption of AI agents.

"The amount of compute necessary for an agent versus a human using it is probably 15 to 100 times depending on the type of problem you're trying to solve."

"Today, the vast majority of AI is prompted by people. I believe that this last month, it has crossed. Most AIs are now agentic."

If Jensen’s claim is right, the compute intensity of the average workload rose by one to two orders of magnitude inside thirty days. And, the demand curve is steepening.

And, it’s not only Jensen saying that agentic adoption is driving compute demand, you can see it across multiple companies.

Intel’s Lip-Bu Tan: “As AI expands from training to inference and increasingly to agentic and multiagent systems, general purpose server CPU density continues to increase, and our core server CPU franchise is growing faster than ever.”

Clarke (COO, Dell) noted agentic workflows "are creating incremental demand for traditional servers", which helped Dell's traditional servers revenue rise by 122% YoY.

Nebius also noted they are "adding ARM and CPU deployments alongside the GPU fleet" for the same reason.

The ROI Question 

The bear case has always been that the returns on AI investments are speculative. Three parties on the sell side of this trade say otherwise.

Jensen Huang relayed what customers are telling him: "return on invested capital is now less than a year. And we're talking about $50 billion data centers." 

Dickerson (Applied Materials) notes that "many of these companies [his customers] are already generating positive returns on their investments." 

And, Tan (Broadcom) talks about the growth of frontier labs,"$30 billion of ARR," which he called "a hell of a business model."

Jensen Huang notes companies are leaning in because they see profitable deployment of compute: "AI is now doing productive and useful work. If we had more compute, we could generate more profitable tokens, which results in more profit for all of the services."

Price For Compute Keeps Rising

Higher returns are driving compute demand, which means compute crunch continues. Buyers are lining up to buy compute at a higher price than before, which helps the entire flywheel. 

Nebius CEO noted they could "sell today our entire 2027 capacity on these terms" and is choosing not to, because retaining capacity for short-term buyers is worth more. 

He expects to get paid better later. This tells you the demand is going higher. 

Supply Constraints Across The Flywheel

The most useful thing about reading these calls together is that every company names a constraint, and no two name the same one. 

But, when you take them together, you find demand is outrunning supply at every level.

At the top of the stack, the limit is clean floor space — highly controlled, ultra-pure real estate inside a semiconductor fab (fabrication facility).

Brice Hill (CFO, Applied Materials) says their growth is driven by availability of "clean room from our perspective". And that clean room availability "will determine what we can all ship next year." 

One layer down it becomes physical inputs, and Tan (Intel) tells the story. The industry faces "one of the most severe supply constraints in its history across leading-edge logic silicon wafers, memory, and substrates," and "these shortages will persist for the foreseeable future." 

Memory was the common bottleneck that showed up across the flywheel. 

Kress (CFO, Nvidia) reset guidance on memory, citing "extreme pricing conditions" where the increase "has exceeded our prior expectations and are headed even higher into next year." 

Her framing: "tighter memory supply is a symptom of the same demand surge that's driving our own growth."

Jeff Clarke (COO, Dell) also gave a shortage list, and it’d make Micron very happy: "DRAM, DRAM, DRAM, followed by NAND, NAND, NAND. We have spotty CPU shortages. There are shortages with disk drives, CDUs, power, racks..."

Intel hired the former CEO of SK Hynix, not saying they were planning to sneak in line by doing this.

The supply constraints don’t stop upstream only. 

Huang: "You've got to go secure the land, power and shell, which oftentimes is a couple 2, 3 years out." 

[ This is one of the reasons we love IPPs here like Vistra, Constellation Energy, and Talen - they are cheap with strong EPS growth.]

Tan (Broadcom) escalated the same language, calling land, power and shell "more than a big concern" because "it dictates specific timing of when this capacity gets deployed." 

It is why he separates contracted gigawatts from deployed ones, judging "conservatively" that less than 30 gigawatts actually goes into production across 2027 and 2028.

Huang gave the best summary, saying the entire chain is stretched, Nvidia currently has supply for roughly 70% of demand, and that he expects "supply to remain a bottleneck at least through the end of fiscal year '28."

The Buyer Base

Markets look at the semis buyer base as 5 hyperscalers, one of whom can go insolvent.

Nvidia’s non-hyperscaler segment covering sovereigns, regional neoclouds, enterprise and edge did $40 billion, grew 138% YoY. Kress expects it to reach "roughly half of our data center business." 

Huang argues the market cannot see it because these buyers "don't buy chips one at a time. They really need an entire factory platform built for them."

His sovereign argument is the structurally strongest point he made, and it connects directly to the bottleneck above: "A country or region can allocate land and power directly to a regional cloud partner in ways it never would to a foreign hyperscaler." 

When land and power are the scarce inputs, the buyer who controls it wins allocation.

Sovereign revenue grew 35% sequentially and more than tripled YoY. There is no better customer in the world than Governments.

Dell also mentions the breadth and growth of buyers. It took Dell eight quarters to reach the first 3,200 AI Factory customers and only three quarters to add the next 3,300. That curve does not come from five hyperscalers.

There are good pickings out there.

I expect that, looking out 12 months from now, datacenter and semi linked names are up 25% to 40% from current levels.

We added Sterling Infrastructure (STRL) and Comfort Systems USA (FIX) this week.

FIX is also part of the Lumida mean reversion strategy which you can find on the Lumida Invest app. The strategy bought it at the close of Aug 26th, and the stock has had 6 out of 7 green days since then.

This shows how AI gives us an edge.

You can also explore Lumida strategies on the Lumida Invest app.

We are working to make these subscribe-able by year-end with direct brokerage integrations. The strategies are ideal for non-taxable accounts as these have higher turnover.

Each strategy page shows its performance compared to SPY, and other relevant risk metrics. You can also see what each strategy has bought or sold and a full trade history. 

On the Lumida Invest app, we have also launched an infinite feed that provides curated insights from high-quality proprietary sources. It’s one way to get an edge against the everyday investor.

We also updated the ‘News Desk’. You can start your morning with a collection of the most insightful and viral articles - and what that means for sectors and securities.

Download the app today to get a glimpse into future of investing.

Sterling builds the shell.

Sterling's (STRL) E-Infrastructure business does site development for large data centers: the earthwork, grading, foundations and drainage that has to be finished before anyone pours a slab or sets a transformer. 

It is the least glamorous line item in a $50 billion facility and one of the earliest on the critical path.

The fundamentals have been compounding hard. 

Revenue is estimated to grow 65% and EPS 84% this year. Return on equity is 40.0%.

And the multiple went the other way.

Forward P/E peaked above 50x earlier this year and now sits at 21.7x. That is a 57% compression in the multiple. PEG is 0.54.

The chart worth looking at twice is the estimate versus the price.

The blue line is NTM EPS. It has stepped higher without interruption and now sits at $22.37. The black line is the share price, which peaked mid-year and has fallen more than 50% over three months.

Earnings up. Price down. It is what dislocation looks like.

The stock closed Friday at $486.49, up 5.75% on the day, which suggests we are not the only ones who noticed.

Comfort Systems keeps it running.

Comfort Systems (FIX) is the mechanical and electrical contractor. It builds HVAC, cooling systems, electrical fit-out, and increasingly, modular assemblies.

More simply, FIX makes the CDUs (coolant distribution units) in Clarke's shortage list.

It is also the part of a data center that cannot be value-engineered away, because a rack of Blackwells that cannot be cooled is a very expensive space heater. 

Revenue is estimated to grow 43% in 2026 with EPS growth projected at 70%. 

Return on equity is 55.3%, with a PEG of 0.67.

Forward P/E has come down from a peak near 47x in April to 30.3x, while earnings have gone higher. The gap between them is the opportunity.

Talking about buys, we also added Genius Sports (GENI) this week.

Every regulated bet placed on a live sporting event has to settle against some evidence – somebody has to say that the pass was completed, the goal stood, and the clock read what it read.

Genius Sports is that somebody for a large part of world sport, including the NFL, whose official data rights it holds exclusively.

Genius collects official league data, distributes it to over 500 sportsbook brands, and prices the odds for different markets those books run. 

And now, after the Legend acquisition, it also owns the audience those books recruit customers from.

Genius Sports’ moat is the exclusive media rights sold league by league. 

When Genius wins a league contract, every book taking bets on that sport has to buy from Genius, because as Mark Locke (CEO, Genius Sports) puts it, "on the sports that we hold, settlement runs on our data." 

A competitor cannot build a better product and take share. It has to wait for the contract to expire and outbid the incumbent for the league itself.

The latest quarter was clean. Revenue of $196 million, up 65%, with betting up 28% and EBITDA of $53 million against $45 million guided. 

Full-year guidance was raised to roughly $1.0 billion of revenue (+51% YoY).

After quarter end, Genius signed direct agreements with both Kalshi and Polymarket, and it gets paid by them three separate times over.

First, the platforms need official data to settle their contracts, same as any sportsbook. 

Second, the market makers who provide liquidity on those exchanges need pricing models to quote against, and Genius has been selling those for twenty years. 

Third, and this is the leg nobody was modelling, the platforms need customers.

That last one comes from Legend, the sports media business Genius bought in May, which owns the sites where bettors are recruited. 

Kalshi and Polymarket are now competing against the sportsbooks for the same users, and Locke says the cost of reaching them is climbing: "there is a battle and clearly, that's causing the premium space to be elevated in price."

Genius sells picks and shovels to both sides of a bidding war it never has to enter. The CEO’s summary: "Our role in this market is infrastructure. We supply everyone."

The valuation is the most interesting case here.

Genius is worth about $2.0 billion. Sportradar (SRAD), its closest comp, has twice the market cap, at a 3x higher P/E valuation. 

Kalshi and Polymarket, its two newest customers, have each been privately valued at several times that.

Forward P/E is 8.9x, at lows relative to history. Management guides to $145 million of unlevered free cash flow in the second half at 70% conversion. That’s a projected FCF yield of about 14% annualized.

Here’s the fundamental-to-price chart. Notice how massive the spread has become- it would have to converge, and that’s the opportunity.

What else did we add?

First Solar (FSLR) - every gigawatt in the AI buildout needs power that can be connected quickly.

And, utility-scale solar is the fastest generation asset to permit and stand up in the United States.

First Solar sells into exactly that demand, and it is the only major panel maker not dependent on a Chinese supply chain.

The stock trades at 9.8x forward earnings, lowest levels of 5Y history, with a PEG of 0.46.

FCF yield is also at 5Y high of 6.8%.

Earnings are expected to grow above 20% this year. However, revenue is guided slightly negative, due to a project timing issue.

The second catalyst for FSLR is rates coming down. FSLR has negative sensitivity to rates, with a beta of -1.29.

US10Y has likely peaked here, and as it comes down, FSLR’s stock performance should improve.

Intuitively, this makes sense. Solar projects are financed assets, so their economics move directly with the cost of capital.

You are paying single-digit earnings for a domestic power supplier to the AI buildout, with a rate decrease as a free option on top.

The best argument for semiconductors comes from the following Renmac positioning chart.

Short-term momentum in Tech just closed in the 1st percentile of its history going back to 1985. Industrials did the same thing in the same week.

This tells you how far the positioning in tech momentum has dropped. Everyone, who was looking to get out, has gotten out. 

This is where semis get interesting. 

Extreme readings tend to mean-revert, and the reversion tends to be sharp because the marginal seller is already gone.

We wrote about the Leopold liquidation as the clearing event. 

The leverage came out, the Korean ETF boom unwound, and retail net buying collapsed. That process has now run its course, and what is left is a category trading on fundamentals again.

We picked up Micron this week.

The logic is simple. It is American-led memory, and the US government now holds an equity stake in the company. 

When Washington is on the cap table, there are things it can do on trade policy with South Korea that it could not do before.

That is a policy option you are getting for free inside the multiple.

The rest of the memory complex is interesting too.

SK Hynix has been trading at cyclical multiples for a business sitting at the center of the AI memory bottleneck.

And the picks-and-shovels names into memory, Applied Materials in particular, have not re-rated with the underlying.

With the positioning unwound, and sentiment reset around semis, the names are starting to look good.

Hyperscalers are pushing harder into custom silicon that competes against Nvidia’s chips. 

Anthropic is now the largest Broadcom customer. OpenAI has secured its own custom silicon program for Jalapeno chips. 

AMD has opted for a safer answer with its deal with Broadcom on custom ASICs. 

What does Nvidia do?

Invest in neoclouds and back open-weight models.

This is feint and counter-feint at the highest level.

I have not seen Jensen make a move yet that I can look at and call bad.

Look at what Nvidia’s investments actually do.

Nvidia's largest customers are now the companies building the chips designed to replace Nvidia. That is customer concentration risk sitting directly on top of competitive risk.

Funding the neoclouds de-risks both at once. It creates demand that does not depend on the hyperscalers, from buyers who have no ambition to build their own silicon.

Backing open-weight models does the same thing one layer up. If inference commoditizes across a thousand deployers instead of five, Nvidia sells to a thousand buyers.

These are the Wintel wars of the 1990s with a larger board and faster clocks.

Competition is the modern form of warfare. It is also the civilized form.

Two things I think are constructive for the tape and are not being priced.

The first is Grok Bot.

The second is the Anthropic S-1, expected Tuesday.

The S-1 is the bigger one, and not because of the IPO itself.

It would be the first time the market gets audited financials from a frontier lab.

Anthropic is reportedly generating meaningfully more revenue than OpenAI. 

Anthropic has also been holding back its next model. My read is they drop it around the process.

I expect Anthropic’s margins are higher - less subsidies than people think.

We are in the month that has rarely worked for markets.

Going back to 1928, September is the only month with a negative average return. 

It is down 1.08% on average, against gains everywhere else, and it is the only month where down months outnumber up months, 53 to 44. 

When September falls it falls harder than any other month, with an average decline of 4.70%.

September in mid-term years also has a reputation. 

Look at the table below, which runs the S&P 500's eleven sectors through September of every mid-term year since 1927.

Ten of the eleven sectors show a negative cumulative return across Septembers in mid-term years. The only survivor is Telecom, at plus 2.9%.

If you look at cumulative numbers, the direction becomes clearer.

Technology is down 82.6% cumulatively, Materials 79.7%, Industrials 69.0%, Real Estate 85.1%. 

Win rates cluster in a narrow band from 39.4% for Real Estate to 55.6% for Telecom, which is to say September is close to a coin flip that pays you badly when it lands wrong.

The average sector loses somewhere between 0.6% and 1.6% during September.

The good thing?

The market already knows it.

Short interest as % of market cap has been increasing for S&P 500, making its yearly highs. This tells you investors are expecting volatility and are positioned for it. 

Also, it’s important to see what a bad September sets up for. 

When September is a bad month, it tends to create the buying opportunity for a year-end rally that usually begins in October. 

Look at the right side of the chart.

October averages 0.53%, November 1.01%, and December 1.26%. December also has the highest count of up months in the entire year at 70.

A bad September isn’t always bad news. 

Arm CEO Rene Haas explains why data center construction, chip production, and memory shortages are limiting AI’s growth even as demand remains insatiable.

View Curation

Sam Altman explains how public market pressure could complicate AI safety decisions, making OpenAI’s mission more important than an IPO timeline.

View Curation

Gavin Baker argues that the data center buildout is creating well-paid trade jobs and boosting local tax revenues, delivering benefits the AI industry has struggled to communicate.

View Curation

Why Trump Won't Escalate in Iran 

Follow Lumida Memes for more.

Not Subscribed Yet? Don’t miss out on future insights—subscribe to the newsletter now!

For real-time updates, follow us on: 

X | Telegram | Youtube | TikTok | News | Ram’s X | Lumida Health | Lumida Tax | Lumida Memes

As Featured In

Disclaimer: Lumida Wealth Management LLC (‘Lumida”) is located in New York, NY, and is an SEC registered investment adviser. SEC registration does not constitute an endorsement of the firm by the Commission nor does it indicate that the adviser has attained a particular level of skill or ability. Lumida only transact business in those states in which it is registered, or qualifies for an exemption or exclusion from registration requirements. Any direct communication by Lumida with a prospective client will be conducted by a representative that is either registered or qualifies for an exemption or exclusion from registration in the state where the prospective client resides.

The information in this material has been obtained from sources believed to be reliable. While all reasonable care has been taken to ensure that the facts stated in this material are accurate and that the forecasts, opinions and expectations contained herein are fair and reasonable, Lumida, Inc. and Lumida Wealth Management LLC (collectively Lumida) make no representations or warranties whatsoever the completeness or accuracy of the material provided, except with respect to any disclosures relative to Lumida. Accordingly, no reliance should be placed on the accuracy, fairness or completeness of the information contained in this material. Any data discrepancies in this material could be the result of different calculations and/or adjustments. Lumida accepts no liability whatsoever for any loss arising from any use of this material or its contents, and neither Lumida nor any of its respective directors, officers or employees, shall be in any way responsible for the contents hereof, apart from the liabilities and responsibilities that may be imposed on them by the relevant regulatory authority in the jurisdiction in question, or the regulatory regime thereunder. Opinions,forecasts or projections contained in this material represent Lumida’s current opinions or judgment as of the day of the material only and are therefore subject to change without notice. Periodic updates may be provided on companies/industries based on company-specific developments or announcements, market conditions or any other publicly available information. There can be no assurance that future results or events will be consistent with any such opinions, forecasts or projections, which represent only one possible outcome. Furthermore, such opinions, forecasts or projections are subject to certain risks, uncertainties and assumptions that have not been verified, and future actual results or events could differ materially. The value of, or income from, any investments referred to in this material may fluctuate and/or be affected by changes in exchange rates. All pricing is indicative as of the close of market for the securities discussed, unless otherwise stated. Past performance is not indicative of future results. Accordingly, investors may receive back less than originally invested. This material is not intended as an offer or solicitation for the purchase or sale of any financial instrument. The opinions and recommendations herein do not take into account individual client circumstances, objectives, or needs and are not intended as recommendations of particular securities, financial instruments or strategies to particular clients. The recipients of this material must make their own independent decisions regarding any securities or financial instruments mentioned herein and should seek advice from such independent financial, legal, tax or other adviser as they deem necessary. Lumida may trade as a principal on the basis of its views and research, and it may also engage in transactions for its own account or for its clients’ accounts in a manner inconsistent with the views taken in this material, and Lumida is under no obligation to ensure that such other communication is brought to the attention of any recipient of this material. Others within Lumida may take views that are inconsistent with those taken in this material. Employees of Lumida not involved in the preparation of this material may have investments in the financial instruments or securities (or derivatives of such financial instruments or securities) mentioned in this material and may trade them in ways different from those discussed in this material. This material is not an advertisement for or marketing of any issuer, its products or services, or its securities in any jurisdiction.