Over on the Left Coast, the turmoil in the New York markets certainly feels a long way away.

In fact, given what’s happening elsewhere, the mood today at one of Silicon Valley’s top financial conferences seems almost complacent (no doubt it owes something to the location, on a remote bluff overlooking the foggy Pacific Ocean.)

And why not? Venture capital worldwide is likely to top $40bn this year for the first time since 2001, and IPO stocks in the US are up 20 per cent since June, compared with a 1 per cent  rise in the Nasdaq. That lubricates a lot of new deals for the financiers and entrepreneurs mingling here.

So it’s a good thing that Paul Wick of J&W Seligman, one of the biggest specialist tech mutual fund investors, was on hand today to cast a cold eye over the good and bad of what’s been happening in the IPO market.

Asked whether he thought any investment banks led better IPO deals than others, he certainly did not shirk the question:

"UBS has underwritten a lot of clunkers," he said, starting with but not limited to Vonage. Then there’s Needham, which "has underwritten lots of crummy companies - [though] they have also on occasion had something everyone has overlooked that has been a diamond in the rough."

Even Goldman Sachs, which he generally credits with backing good companies, does not escape the scorn. "Glu Mobile was a real clunker," he said of the mobile games company whose shares were priced by Goldman at $11.50 in an IPO earlier this year and now change hands for around $5. But then, that should have been an easy one for investors to spot: "The Glu Mobile CEO previously rode 3dfx [Interactive] into the ground."

Then there are the sell-side analysts, who were meant to have cleaned up their act after the scandals of the dotcom bubble.

"The sell-side has gotten worse and worse over the years in sucking up to the VCs and sucking up to the public companies," according to Wick. "It’s really quite disturing." He points to coverage of companies like Data Domain, whose rising stock prices seem to be an excuse for analysts to simply raise their price targets higher.

"As you get close to the 6 month lock-up date [for Data Domain], suddently everyone’s target prices have crept up," he complains. "It’s really bothersome for those of us trying to navigate the public markets. What we see happening, the sell-side totally ignores the fact that there’s a ton of competition coming at them in 2008, a lock-up expiration is coming. We’re supposed to just hang on to these things - and buy more."

For good measure, Lise Buyer, a former sell-side analyst who masterminded Google’s IPO, had this to say about the latest crop of new internet companies: "Putting a double vowel in the name of a company doesnt actually guarantee a better valuation." You have been warned.

It seems that BEA Systems can juggle numbers just as much as Oracle can (see note below.)

Announcing earnings on Thursday, CEO Alfred Chuang wanted Wall Street to overlook his gigantic stock option backdating charge (which virtually wiped out reported profits for the past decade) and focus instead on the software company’s rebounding profit margins. And not just any profit margins but pro forma figures which, among other things, excluded the costs of hiring all those lawyers and bankers to keep the barbarians (in the form of Larry Ellison and Carl Icahn) from the gates.

Shouldn’t expenses like these be seen as a normal cost of doing business, one analyst wondered?

It isn’t normal to come under fire from a hostile bidder and a shareholder activist at the same time, retorted BEA executive Bill Klein. "We don’t believe that both of these circumstances can continue for the long run." Is that reasoned analysis or wishful thinking?

Growing a company through acquisitions can feel a bit liking eating Chinese food. Two hours after you’ve had a meal you’re hungry again.

Larry Ellison must be starting to feel that way. When he bought PeopleSoft, he promised it would put Oracle on course for earnings per share growth of at least 20 per cent a year. He’s more than met that promise (that is, if you care to overlook the little matter of amortisation - more on that below.)

The trouble with strategies like this is, they only work if you keep buying. Prices have been going up and suitable big targets are getting scarcer - which explains why this has now become a preoccupation on Wall Street, to judge by the questions Ellison faced at his company’s annual financial analyst meeting on Wednesday.

Ellison’s message to the doubters: "We have found some very attractive targets, and we think we can do it for a while longer." Compared to the horizontal software plays he has bought so far, the next spate of deals might involve more "vertical" companies that specialise in specific industry sectors. "We can buy a typical vertical company that’s running at 5 per cent [operating margins] and run it at 30 or 40 per cent," Ellison boasted.

The trouble is, the bigger Oracle gets, the hungrier it becomes. According to Charles Di Bona at Sanford C Bernstein, Oracle will have to spend $52-58bn on acquisitions over the next five years to keep its earnings per share machine ticking over at the desired rate. So either the deals have to get bigger - in which case the risk of overpaying and the complexity of integration go up - or Oracle has to spread its net wide and trawl for large numbers of smaller transactions.

There’s no question that Ellison’s consolidation play has been a huge success. The outcome of his current circling of the beleagured BEA Systems should give some idea of how long he can keep it up.

Footnote: What is the right measure of earnings for a serial acquirer like this, anyway? Oracle likes to boast about its pro-forma earnings per share. But as the deals have flowed, the amortisation of intangibles has picked up: 12 per cent of operating profits in the 2006 fiscal year, rising to 15 per cent in 2007.

Look at it another way. Back in 2004, Oracle’s earnings per share looked roughly the same on both a GAAP and non-GAAP basis. In the three years since, the pro-forma number has jumped by 98 per cent: the GAAP number is only up 62 per cent. Like too much Chinese food, all those deals start to show on the waistline in the end.

Another day, another 3 per cent on Baidu’s share price. The Chinese search engine is now deemed to be worth more than $12bn, which is double what the stock market thought it was worth as recently as August. Not to be left behind, portals like Sohu and Sina are on a run of their own, with gains respectively of 160 and 130 per cent this year.

For Baidu, if you’re keeping score, that puts the company on a multiple of more than 30 times next year’s revenues, and 94 times 2008 earnings.

With nearly a year to go until the Beijing Olympics, what’s the chance that this bubble will keep inflating at least that long? No surprise that the Olympics were on the lips of a senior Sohu executive when the company reported earnings today, and are contributing to a general belief that this rally is bullet-proof.

The long-run potential for the companies that end up dominating the Chinese internet sector may indeed be huge, but the US internet bubble showed how hard it is for investors to discount that back to a realistic valuation for today’s nascent internet giants - why should this be any different?

Of course, it doesn’t hurt Baidu’s cause if its market share is being inflated in unusual ways (see this report on how internet users who try to visit Google end up on Baidu.) As Sergey Brin groused last week: "Obviously that makes it very hard to do business, when your customers are redirected to a competitor."

The pain from the long dotcom hangover is finally starting to recede into the past, at least when it comes to venture capital returns. About time, too. But it is still far too soon to tell whether historic long-term profits from VC investment will hold up.

The story is told in the chart below. The thick broken line at the bottom shows five-year venture capital returns in the US. As write-offs from the dotcom disaster have receded and profitable exits are being found for the the companies that survived, this line has finally crept back into positive territory (the latest figures were put out today by the NVCA.)

The most encouraging part of this chart is the thin broken line in the middle: despite the boom and bust, 20-year returns from start-up financing have stayed remarkably solid, at around 16 per cent a year.

But will that continue to hold good for the next 20 years? The top line shows how ten-year returns, which had been boosted by the bubble, are sinking back towards the norm. The supply and demand equation in venture financing looks very different than it did back in the mid-1990s, with many more funds and many more investors still scrambling to get in. That surely points, eventually, to long-term returns below the historic mean.

Great company it may be, but VMware’s share price is starting to exhibit some very bubbly symptoms.

The virtualisation software company’s shares are now trading at more than three times what they were sold for at the time of its IPO less than two months ago. That has lifted its stock market value to $34bn, and raised the value of EMC’s remaining stake in the company to around $30bn.

This is where things start to get weird. EMC’s total market value is only $44bn. Sure, it has been buoyed by the VMware IPO, but to nothing like the extent that would seem logical given what’s happened to the VMware stock.

It is all faintly reminiscent of what happened when 3Com sold a small stake in its red-hot Palm division during the tech boom. Back then Wall Street accorded Palm a valuation well in excess of 3Com, even though 3Com continued to hold most of the stock.

Of course, things righted themselves pretty fast. 3Com rushed through with a spin off of the rest of Palm, the market became flooded with Palm stock, and the price collapsed. Something similar could be in store for VMware shares if EMC decides to distribute a sizeable portion of its holding. A "lock-up" agreement prevents any of the VMware holders from selling until six months after the IPO, so the scarcity premium may last a while longer, but it seems a fair bet that supply will quickly catch up with demand after that.

It looks like, right now, things are running about as smoothly as they could be for tech financings. Wall Street’s IPO window has been open all year. Now the mergers and acquisition market is warming up nicely as well. Ironic, really: the sub-prime debt crisis has done nothing at all to hamper the risky end of the stock investment spectrum.

According to the National Venture Capital Association, M&A activity involving venture-backed companies in the US has just rebounded to its highest level since the beginning of 2001. Valuations are also going up: for transactions where the price was disclosed, the average deal size in the latest quarter reached $226m, the highest since late 2000.

The IPO market, for its part, took a bit of a pause over the summer, but Wall Street’s attention was kept firmly on the tech sector’s charms thanks to the eye-popping debut of VMWare (not included in the NVCA figures.) We mused a while ago about the apparent "underpricing" of VMWare’s IPO, but that turns out to have been an understatement: the virtualisation software company now has a stock market value of $34bn, or $23bn more than the bankers valued it at when they took it public six weeks or so ago. It isn’t hard to find the reasons why VMWare has taken off: a small float of shares, a super-hot company, and a stock market that is hungry for quality growth companies. A perfect storm like this rarely lasts.

With all the exits now open, expect venture capital to keep flowing strongly. This is about as good as it gets for tech financings - just as long as the party doesn’t start to get completely out of hand.

It’s starting to look like the good-old, bad-old days in the tech finance business.

Remember when investment banks stood accused of deliberately under-pricing tech IPOs so they could hand big windfall gains to their best institutional investor friends? It was to counter that sort of thing that Google opted for an auction when it went public three years ago this week.

VMWare’s IPO today smacks of an earlier era. An opening day premium of 76 per cent pushed its stock market value up close to $20bn - and all that in spite of its mid-August launch date and the highly unstable state of the financial markets.

The size of that premium suggests that VMWare’s bankers left $725m on the table by pricing the deal lower than they could have done. That is considerably more, as it happens, than EMC paid when it bought all of VMWare three years ago.

Back during the tech bubble, the bankers said they simply had no control over first-day "pops" like this. If traders wanted to chase prices to irrational levels in pursuit of the instant profits on the latest hot stock, that was no reason to price shares at nonsensical levels.

It’s harder to make that argument today. Demand for VMWare was always going to be heavy: it’s not every day a software company whose revenues have reached an annualised $1bn, growing at nearly 100 per cent, makes its appearance on Wall Street. It would be interesting to know how much more EMC would have made had it opted for a Google-style IPO, rather than leaving it to Citi, JP Morgan and Lehman Brothers to find the "right" level at which to sell the shares.

First, the cold. Limelight Networks was meant to be the New Akamai, which made its June IPO one of the most anticipated of the year. Content delivery networks like this, which bring high levels of reliability to routing data across the internet, should be big beneficiaries of the growth of online video. They are also good barometers of the health of the broader online media industry.

If Limelight executives are to be believed, the rise of online video is turning out to be far from smooth. The company today gave a very downbeat revenue forecast for this quarter, in the process triggering a 35 per cent drop in its share price. Jeff Lunsford, chairman and chief executive officer, blamed it on seasonality: apparently demand for Web video drops off as the regular TV seasons end for the summer. He also pointed out that many online media businesses are still searching for a business model. The costs of delivering online video are soaring before the revenues start to come in, leading media companies to demand lower prices from companies like Limelight.

The issues faced by one relatively small network company may not be typical of the whole industry, and the audience lull and the difficultly of making money from an online audience may be temporary. But this is a reminder that the online video business is still in its infancy.

Now the hot. VMWare’s debut on Wall Street, due next week, looks like being the tech event of the year in the IPO market. The company, which makes virtualisation software, has just raised the indicated price range for its shares from $23-25 to $27-29. At the upper level, that would vaue it at nearly $11bn, putting it comfortably among the world’s top ten software companies by market value.

As we wrote a while back, CEO and co-founder Diane Greene is about to step forward as one of Silicon Valley’s most powerful women.

Intel’s cosy relationship with Apple, supplying the microprocessors for its computer range, does not extend to the iPhone yet. But there may be a way into the hottest cell phone through Intel’s WiMAX technology.

The applications processor in the iPhone is supplied by Samsung and uses a core based on the ARM architecture for small devices, rather than Intel’s x86 architecture, which dominates the PC world.

Sriram Viswanathan, a vice president at Intel Capital, its venture arm, and head of its WiMAX programme, says the ARM processor limits internet access such as unfettered viewing of YouTube videos.

In an interview, he spoke about bringing the x86 world to smaller gadgets - mobile internet devices, as Intel describes them. This will be enabled by its ultra-low power, small form-factor Silverthorne chip, which will eventually have a WiMAX chip attached.

AT&T, the exclusive carrier for the iPhone in the US, is bidding for new spectrum here and could well choose to run a WiMAX next-generation network on it if successful, he says.

Although he did not make the connection, this could open the way for Intel to at least supply a WiMAX chip, and perhaps a Silverthorne microprocessor for future generations of the iPhone.

Intel’s WiMAX chips, which offer broadband connectivity over wide areas, will definitely be featuring alongside regular Wi-Fi chips in new high-end notebooks from the second half of next year - about the time that new WiMAX networks around the world will be going live.

Intel appears to have been behind the deal this month that saw Sprint and Clearwire team up in the US to create a national network, although Mr Viswanathan would only say: “We were very actively involved…it was obvious to us that there was tremendous benefit in them working together.”

Intel owns about 30 per cent of Clearwire, after putting more than $600m into the wireless provider, its biggest ever venture-capital investment.

“We are not investing just enough anymore, we are investing as much as we can to meet our strategic objectives,” he says of Intel Capital’s new tactics to open up fresh markets for the company’s chips.

“If we think we can drive the market, it makes sense to put as much money as we can into that area.”